JP2001285788A - Time sequential information storing/reproducing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a time sequential information storing/reproducing device for performing the storage processing of event data so as to improve retrieval efficiency to the recording medium of a fixed capacity and realizing quick and easy retrieval at the time of retrieving contents later. SOLUTION: A first information storing/reproducing part 5 preserves all new event data first. After the lapse of prescribed time, the event data of the first information storing/reproducing part 5 are integrated/selected and the event data are transferred and recorded in a second information storing/ reproducing part 8 thereafter. The second information storing/reproducing part 8 integrates/selects and then preserves the event data before the certain point of time for a long period of time.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION The present invention extracts only event information to be monitored from event information that is continuously input in a time series regardless of whether or not an event to be monitored has occurred. The present invention relates to a time-series information accumulating / reproducing apparatus for accumulating data.

[0002]

2. Description of the Related Art Event data obtained in a time series for an event occurring in a monitoring area (image data of an event obtained by imaging the monitoring area, audio data of an event obtained by collecting sound from the monitoring area, The following prior art relates to an apparatus for monitoring events that are to be monitored and that uses sensor data to constantly monitor the events to be monitored and that records the events to be monitored in detail.

(1) Event detection sensor and time lapse VT
In combination with R, images, sounds, sensor information, and the like are sequentially recorded at long time intervals when no event occurs, and at short time intervals during event occurrence. (2) An event is detected by image processing, and an image is recorded only when an abnormal event is detected. (JP-A-11-16059)

(3) Further, as a conventional technique which is generally widely implemented, there is a monitoring apparatus as shown in FIG. This monitoring device monitors an event that occurs in a monitoring area. In this monitoring device, 110 is an imaging unit, 120 is an A / D (Analog / Digital) conversion unit, 13
0 is an image memory for storing digitized image data, 140 is compression means for compressing image data, 150 is external signal input means, 160 is image writing means, and 170 is an information storage / reproduction unit. As a specific example of the information storage / playback unit 170, a method of writing image data to a ring buffer on a hard disk is often used. Although not shown, there is a data format including audio data in addition to the image data.

[0005] Further, as a method for summarizing and storing information, which is not used for monitoring, there is the following conventional technology. (4) When the storage period of the information storage unit has passed the predetermined period, or when the free space of the information storage medium has become equal to or less than the predetermined value, the event detection period or the period specified by the user input information The time series information is compressed and stored by changing the compression ratio of the image information and the compression ratio of the image information in other sections. (JP-A-10-191245, JP-A-10-21427)
0) (5) Summarize and record the image by image processing, record the summary image on the file server in a form that can support later retrieval, editing, processing, and quick viewing of the image, and play it back according to the user's request I do. (Japanese Patent Application Laid-Open No. 8-251540)

[0006]

The conventional monitoring method has the following problems. (1) In a system in which a monitoring image recorded on a recording medium such as a tape is reproduced / confirmed later as in the prior art described in Japanese Patent Application Laid-Open No. 8-277080, it takes time and effort to search for an event occurrence portion from the image. Further, the capacity of the recording medium is fixed, and the data exceeding the capacity may be erased by overwriting and cannot be reproduced later.

Further, when data is recorded over a plurality of recording media even if they are not erased, it is necessary to search for the plurality of recording media, which is troublesome. Also, time-lapse VTR
When such a storage medium is used, the judgment is made based on the monitoring result of the present or very close past, and the judgment cannot be made with reference to the event occurrence situation in the past. Even if there is a past event occurrence situation in the memory of one monitor, the memory cannot be shared with another monitor because there is no specific reference image data.

(2) Even in the method of recording images before and after an alarm is generated by some kind of alarm detecting means as in the prior art described in JP-A-11-16059, one V
Since the recording capacity of a recording medium such as a TR or a hard disk is constant, if the recording capacity is exceeded, overwriting must be performed from the beginning, or a backup must be taken to overwrite. Overwriting deletes data before a certain time,
Further, even if a backup is taken, the trouble of searching for events of all time later becomes enormous.

(3) Also, in the monitoring apparatus of the prior art as shown in FIG. 27, the recording medium has only a fixed recording capacity, and after the recording capacity is used up, backup is taken or You can only overwrite the old record,
There is a problem similar to (1) or (2) described above.

(4) In the prior art described in JP-A-10-191245 and JP-A-10-214270, images, sounds,
Alternatively, record images only when an event occurs, or change the image compression ratio or audio compression ratio between when the event is detected and other times, and record only useful information for user input. However, it is assumed that a single type of abnormality or event is detected, and there is only a classification such as normal / abnormal or event detection section / other section.

In actual monitoring over a long period of time, it is desirable to check the general occurrence of events within the monitoring target period before viewing individual situations. In addition, it is necessary to treat a plurality of types of events as a group, such as repeated occurrences of the same event, occurrences having a pattern of a plurality of events, and no event occurring for a long time. But
In this conventional technique, data for grasping the outline of such an event is not accumulated, and it is not possible to cope with the layering of the event.

(5) Even in the prior art described in Japanese Patent Application Laid-Open No. 8-251540, similar to (4), reference data for grasping the outline of the event is not stored, and it is possible to cope with the hierarchy of events. Can not.

In actual monitoring over a long period of time, it is desirable to be able to confirm the general occurrence of events within the monitoring target period, and then to view individual situations. in this case,
Efficient monitoring can be performed by adopting a data storage format conforming to a human memory mechanism.

For example, the human memory mechanism is various depending on the person or the case, while the memory that frequently occurs is well remembered, while the memory that rarely occurs is clearly remembered. However, in general, it is a matter of forgetting unnecessary information, that is, selecting and discarding unnecessary information, which is excellent in determining whether information is important.

By adopting a data storage format such as a human storage mechanism, a hierarchy can be introduced according to the importance of information, and monitoring efficiency is improved. For example, in the near past, the occurrence of each and every event is confirmed, and in the distant past, a rough appearance of the occurrence of an event is captured. However, in these prior arts (3) to (5),
As described above, such improvement in monitoring efficiency could not be expected.

Generally, in the prior arts (1) to (5),
The storage processing is not performed so as to increase the search efficiency of the monitoring target, and the search efficiency is extremely low when the content is searched later. In view of such a situation, there is a need for a time-series information accumulating / reproducing device capable of improving search efficiency.

[0017]

SUMMARY OF THE INVENTION According to the present invention, a storage medium of a fixed capacity is subjected to a process of accumulating event data so as to increase the search efficiency, thereby realizing a quick and easy search when searching for contents later. It is an object of the present invention to provide a time-series information storage / reproducing apparatus for performing the above-mentioned operations.

[0018]

In the present invention, the following means are provided to solve the above-mentioned problems. Specifically, according to the first to tenth aspects of the present invention, after event data is recorded on a recording medium having a fixed capacity, the event data is reconstructed hierarchically. By relocating and recording on another recording medium, a time-series information accumulating / reproducing apparatus which can easily grasp the outline of the event occurrence state of the entire monitoring time at the time of later retrieval.

That is, according to the first aspect of the present invention, there is provided an event data input means for inputting event data relating to an event continuously in time series, a data storage unit for storing definition data of the event, Event detection means for comparing the event data output from the input means with the definition data stored in the data storage unit and detecting the occurrence and end of the event for a plurality of types of events from the events occurring in a time series. When an event is detected by the event detecting means, first index data including addresses relating to the type, occurrence time, end time, and recording position of each event is created to refer to the event data later. Event data and the first
Writing means for controlling to write the index data together with the event data;
A first information storage / reproducing unit which records and accumulates index data on a recording medium, and reproduces desired event data from the recording medium with reference to the first index data after the accumulation;
Event data of a plurality of events or one of the event data integrated as one event group from the event data recorded in the information storage / reproduction unit
Event integration / selection condition setting means for setting event integration / selection conditions for extracting event data of the event as an object to be stored, and event integration / selection conditions set by the event integration / selection condition setting means. Integrate event data of multiple events into event data of one event group, or select event data of one event, and refer to the event or event group for later reference. Event integration / selection means for creating second index data including addresses related to the type, occurrence time, end time, and recording position of a group, and controlling to write the event data and the second index data together; The event data and the second index data transferred from the first information storage / reproduction unit under the control of the integration / selection means are recorded and stored on a recording medium. A second information storage / reproduction unit that reproduces desired event data with reference to the second index data after the storage, and wherein the first information storage / reproduction unit Is characterized by storing all event data newer than a certain point in time, and the second information storage / reproducing unit integrates and selects event data older than a certain point in time and stores it.

The invention described in claim 2 is the same as the invention described in claim 1.
In the time-series information accumulating / reproducing apparatus described above, the event integration / selection condition is a time interval of the same event that occurs twice in succession, and the event integration / selection unit performs a time shorter than this time interval. It is characterized in that a plurality of the same events that occur continuously at intervals are collectively integrated as one event group.

The invention described in claim 3 is the first invention.
Wherein the event integration / selection condition is a period when normal events occur continuously, and the event integration / selection means generates the event beyond this period. It is characterized in that events during normal times are integrated as one event group.

The invention described in claim 4 is the first invention.
Wherein the event integration / selection condition is the number of times that the same multiple events repeatedly occur within a predetermined time interval or less, and the event integration / selection means When the number of occurrences of the event is larger than the number of times, the plurality of identical events are collectively integrated as one event group.

The invention described in claim 5 is the first invention.
Wherein the event integration / selection condition is the number of times when a plurality of different events occur alternately within a predetermined time interval or less, and the event integration / selection means When the number of occurrences of the event is larger than the number of times, the plurality of identical events are collectively integrated as one event group.

The invention described in claim 6 is the first invention.
In the time-series information accumulating / reproducing apparatus described in the above item, the event integration / selection condition is whether or not a predetermined event occurrence pattern is matched, and the event integration / selection means It is characterized in that events in the case of matching a pattern are collectively integrated as one event group.

The invention described in claim 7 is the first invention.
Wherein the event integration / selection condition is a predetermined accumulated time obtained by accumulating a plurality of times at which the same event occurs within a predetermined period, and the event integration / selection condition is
The selecting means selects an event occurring over a period of time equal to or longer than the predetermined accumulated time or an event occurring over a period shorter than the predetermined accumulated time from a plurality of events.

The invention described in claim 8 is the first invention.
Wherein the event integration / selection condition is a predetermined number of occurrences of an event within a predetermined period, and the event integration / selection means selects the predetermined event from a plurality of events. It is characterized in that an event occurring more than the number of occurrences or an event occurring less than the predetermined number of occurrences is selected.

The invention according to claim 9 is the first invention.
Wherein the event integration / selection condition is a predetermined occurrence frequency of a certain event within a predetermined period, and the event integration / selection means selects the predetermined event from a plurality of events. An event that occurs at a frequency higher than the occurrence frequency or an event that occurs at a frequency lower than a predetermined occurrence frequency is selected.

According to a tenth aspect of the present invention, in the time-series information accumulating / reproducing apparatus according to the first aspect, the event integration / selection condition is a predetermined average occurrence period of an event within a predetermined period. The event integrating / selecting means selects an event occurring in a period longer than the predetermined average period or an event occurring in a period shorter than the predetermined average period from a plurality of events.

According to the eleventh aspect of the present invention, there is provided a time-series information accumulating / reproducing apparatus for selecting and reproducing an event or a group of events so as to enhance monitoring efficiency during reproduction. I do.

That is, according to an eleventh aspect of the present invention, in the time-series information accumulating / reproducing apparatus according to any one of the first to tenth aspects, a reproduction event for an event group or an event to be reproduced is provided. Playback event condition setting means for specifying conditions; event data selection means for selecting event data relating to an event group or an event corresponding to the playback event condition; and a first information storage / reproduction unit and (2) Event data reproducing means for reproducing from the information storing / reproducing unit, wherein the reproducing event condition for the event group or event to be reproduced includes all event data stored by the first information storing / reproducing unit; Further, it is characterized in that part or all of the integrated / selected event data stored by the second information storage / playback unit is stored.

Further, according to the twelfth aspect of the present invention, there is provided a time-series information storage / reproducing apparatus which can store information in a hierarchical manner and makes it easier to search later.

That is, according to a twelfth aspect of the present invention, in the time series information storage / reproduction device according to any one of the first to eleventh aspects, the second information storage / reproduction unit is provided at a subsequent stage. New event integration / selection condition setting means, event integration /
It is characterized in that the number of layers when event data is stored is increased by using K (K is a natural number) stages of information storage / playback units by cascading the selection means and the information storage / playback units. .

According to the thirteenth aspect of the present invention, information storage similar to a human memory mechanism (for example, an event that frequently occurs is forgotten as having low importance and rarely occurs) Is an information accumulation that memorizes that the importance is high, and an event that frequently occurs is stored as a high importance, and an event that rarely occurs forgets that the importance is low.) A time-series information accumulating / reproducing apparatus that automatically performs the monitoring without human intervention, thereby lowering the monitoring cost and realizing the same monitoring performance as monitoring by a human. In addition, even if the observer changes, the stored image / sound, etc., of the event before the change is reproduced and grasped in a short time, and the time-series information storage / reproduction device realizes the continuous monitoring. .

That is, the invention according to claim 13 is an event data input means for inputting event data relating to an event continuously in time series, and data for storing event definition data and occurrence frequency data of the event. The storage unit compares the event data output from the event data input unit with the definition data stored in the data storage unit to determine the occurrence and end of an event from a plurality of types of events occurring in time series. Event detection means for detecting an event, and, when the occurrence of the event is detected by the event detection means, an address relating to the type, occurrence time, end time, and recording position of each event in order to refer to the event data later. Writing means for creating index data, and controlling the writing of both the event data and the index data. Record storing the data and index data to a recording medium, after the storage information storing and for reproducing from the recording medium the desired event data referring to the index data
A reproduction unit; an event occurrence frequency calculation unit configured to calculate occurrence frequency of a certain event within a certain period using index data accumulated in the information accumulation / reproduction unit to generate occurrence frequency data; An event occurrence frequency updating means for overwriting and updating the occurrence frequency data generated by the means with the occurrence frequency data in the data storage unit, and reading the occurrence frequency data from the data storage unit and calculating the importance of each event in the period And calculating the event data on the recording medium of the information storage / reproducing unit according to the degree of importance calculated by the importance calculating means. An event data sorting means for rearranging and recording in a continuous recording area in a time series information storage / reproduction apparatus, Reproducing unit, and updates by overwriting the event data newly inputted in a continuous area on the less important event data based on the importance degrees calculated by the importance calculating means.

As for the event data input means common to the first and second inventions, a more specific solution means will be presented in claims 14 and 15.

That is, according to a fourteenth aspect of the present invention, in the time-series information storage / reproduction device according to any one of the first to thirteenth aspects, the event data input means is configured to collect sound by voice. Sound collecting means for outputting a signal, imaging means for outputting an image signal by imaging, or sensing means for outputting a continuously generated sensor signal;
Image signal or sensor signal is converted to audio data,
A / D conversion means for converting the data into image data or sensor data.

According to a fifteenth aspect of the present invention, there is provided the time-series information storage and storage device according to any one of the first to thirteenth aspects.
In the reproducing apparatus, the event data input unit is a receiving unit that receives and outputs audio data, image data, or sensor data transmitted through a communication facility.

As for the event detecting means common to the first invention and the second invention, the solving means is more specifically presented in claims 16 to 21.

That is, according to the invention of claim 16, in the time-series information storage / reproduction device of claim 14 or 15, the event data output by the event data input means is at least image data of the event. The data storage unit stores data relating to a change or a change distribution as definition data, and the event detection unit includes an image event detection unit that detects a change or a change distribution occurring in image data. Features.

According to a seventeenth aspect of the present invention, in the time series information storage / reproduction device according to the fourteenth or fifteenth aspect, the event data output from the event data input means is at least image data of the event. The data storage unit stores a specific motion pattern or a motion pattern distribution generated as definition data, and the event detection unit detects a specific motion pattern or a motion pattern distribution generated in image data. It is characterized by comprising image event detecting means.

According to a twelfth aspect of the present invention, in the time series information storage / reproduction device according to the fourteenth or fifteenth aspect, the event data output by the event data input means is at least image data of the event. Including, the data storage unit stores a specific pattern or pattern distribution in the image data as definition data, the event detection means,
An image event detecting means for detecting a specific pattern or a pattern distribution in the image data is provided.

According to a nineteenth aspect of the present invention, in the time series information storage / reproduction device according to the fourteenth or fifteenth aspect, the event data output by the event data input means is at least audio data of the event. Wherein the data storage unit stores a change in voice data as definition data, and the event detection unit includes a voice event detection unit that detects a change in voice data.

According to a twentieth aspect of the present invention, in the time series information storage / reproduction device according to the fourteenth or fifteenth aspect, the event data output by the event data input means is at least audio data of the event. Wherein the data storage unit stores a specific pattern on audio data as definition data, and the event detection unit includes an audio event detection unit for detecting the specific pattern on the audio data.

According to a twenty-first aspect of the present invention, in the time series information storage / reproduction device according to the fourteenth or fifteenth aspect, the event data output by the event data input means is at least sensor data of the event. Wherein the data storage unit stores a specific pattern on the sensor data as definition data, and the event detection unit includes an event direct detection unit for detecting the specific pattern on the sensor data.

[0045]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing a first embodiment of the present invention;
The twelfth to twelfth aspects will be collectively described as a first embodiment. FIG. 1 shows the time-series information storage and
It is a block diagram of a playback device. The time-series information accumulating / reproducing apparatus according to the present embodiment includes an event data input unit 1, a data storage unit 2, an event detecting unit 3, a writing unit 4, a first information accumulating / reproducing unit 5, It comprises at least a selection condition setting means 6, an event integration / selection means 7, and a second information storage / reproduction unit 8. Further, the first storage capacity monitoring means 9
And second storage capacity monitoring means 10 and time management means 11
And a monitoring start input unit 12.

Further, the event detecting means 3 includes a plurality of event detecting means (not shown) for detecting individual events (not shown), which operate in parallel to detect the occurrence and end of each event. The writing means 4 includes an event data writing control means 4a and an index data creation means 4b.

The event data input means 1 receives event data relating to events occurring in the monitoring area continuously in a time-series manner. The data storage unit 2 stores definition data of an event to be monitored. The event detection means 3 stores the event data output from the event data input means 1 and the data storage unit 2
And detects the occurrence and end of the monitoring target event from among the events occurring in time series for a plurality of types of events, and outputs an event detection result signal.

When the event detection unit 3 detects an event to be monitored, the writing unit 4 refers to the event data so as to refer to the event data later.
First index data including an address related to the recording position of the information storage / reproduction unit 5 is created, and the first information storage / reproduction unit 5 is controlled so that both the event data and the first index data are written.

The first information storage / reproduction unit 5 records and stores event data and first index data on a recording medium under the control of the writing unit 4. The recording medium has a ring buffer structure, in which event data is recorded in the ring buffer, and first index data is recorded in the first index data storage. After the accumulation, the first information accumulation / reproduction unit 5 can reproduce desired event data from the recording medium with reference to the first index data. Here first
The information storage / reproduction unit 5 includes, for example, a hard disk,
It is a large-capacity RAM (Random Access Memory) or the like.

The event integration / selection condition setting means 6 selects the second event data from the event data recorded in the first information storage / reproduction unit 5.
In order to select event data to be moved to and stored in the information storage / reproduction unit 8, a plurality of events integrated as one event group or event integration / selection conditions for determining one selected event are set. .

The event integration / selection means 7 selects an event group to be integrated according to the event integration / selection condition set by the event integration / selection condition setting means 6 or event data relating to the selected event. Furthermore, in order to efficiently refer to these event data later, second index data including the type of the event or event group, the occurrence time, the end time, and the address related to the recording position of the second information storage / reproduction unit 8 is newly added. Then, control is performed so that the event data and the second index data are written together with the recording medium of the second information storage / reproduction unit 8.

The second information storage / reproduction unit 8 stores the event data transferred from the first information storage / reproduction unit 5 together with the second index data on a recording medium under the control of the event integration / selection means 7. The recording medium has a ring buffer structure, in which event data is recorded in the ring buffer, and second index data is recorded in the second index data storage unit. After accumulation, desired event data can be reproduced with reference to the second index data.

With such a time series information accumulating / reproducing apparatus, the event integrating / selecting means 7 is set based on the event integrating / selecting condition set by the event integrating / selecting condition setting means 6.
Is used to integrate a plurality of related events from the event data of a plurality of events recorded in the first information storage / reproduction unit 5 as event data of one event group, and Event data of one event group is recorded so as to be in a continuous area of the recording medium, and event data relating to one event is selected and recorded on the recording medium of the second information storage / reproduction unit 8. Data can be recorded.

Further, the first information storage / reproduction unit 5 includes first storage capacity monitoring means 9 for monitoring the recording capacity of the recording medium. Similarly, the second information storage / reproduction unit 8 controls the recording capacity of the recording medium. There is provided a second storage capacity monitoring means 10 for monitoring. With this configuration, the event integration / selection is set when the recording capacity of the recording medium of the first information storage / reproduction unit 5 and the second information storage / reproduction unit 8 falls below a predetermined capacity. can do. Note that even if the first storage capacity monitoring means 9 and the second storage capacity monitoring means 10 are not provided, the time management means 11 may perform event integration / selection every time a predetermined period elapses. As described above, various times can be selected for the timing of event integration / selection.

Writing means 4 and event integrating / selecting means 7
Is time-managed by the time management means 11, and outputs an occurrence time and an end time when an occurrence and an end of an event are detected. The time management means 11 starts time management from when the monitoring start is instructed by the monitoring start input means 12.

Next, the information storage operation of this embodiment will be described. First, an information storage operation is started by input to the monitoring start input means 12, and time-series event data is sequentially input to the event data input means 1. When the start of monitoring is instructed by the monitoring start input means 12, the recording media of the first information storage / reproduction unit 5 and the second information storage / reproduction unit 8 are all cleared once, and the event data and the first and second information storage units are cleared.
All index data is deleted.

The event data input means 1 includes the event detection means 3
And the event data to the writing means 4 respectively. The event detection means 3 determines whether or not a monitoring event has occurred in event data relating to an event occurring in the monitoring area output from the event data input means 1. The resulting event detection result signal is sent to the writing means 4. The writing means 4 stores, based on the event detection result signal and the input event data, event data in a time zone in which the event occurs or in an extended time zone including before and after this time zone, as first information storage / reproduction. Recorded in the unit 5 together with the first index data. In the recording medium of the first information storage / reproduction unit 5, the first storage capacity monitoring unit 9 checks the data storage capacity by writing event data.

FIG. 2 is an explanatory diagram showing an example of a process of storing time-series event data in the first information storage / reproduction unit 5. FIG. 2A shows event data and an event detection result signal generated in time series and output from the event data input unit 1. Here, the event detection result signal includes the type of event, the occurrence time, and the end time, and two types of event occurrences of events A and B are detected for the event data within this period T. Only the event data in the section in which is generated is extracted. Then, the first index data is generated by the index data creating means 4b based on the event detection result signal, and the first index data is generated by the event data writing control means 4b in FIG.
Are stored in the first information storage / reproducing unit 5 as shown in FIG.

Here, the first index data has at least an event occurrence time, an event end time, an event type, and a recording position of the event data. In addition, in order to facilitate later retrieval, FIG. The event ID as shown in Fig. 4 is attached.
In this case, for the events other than the events A and B to be monitored (that is, stored as definition data in the data storage unit 2), the first information accumulation / reproduction unit 5 determines that no event has occurred. Is discarded without being stored. In this way, the first information storage / reproducing unit 5 stores event data.

When a certain period of time has elapsed since the start of monitoring, a certain period of time has elapsed since the previous event integration / selection processing, or the remaining recording capacity of the recording medium has been reduced, the event integration / selection condition setting means In accordance with the settings made by 6, the event integrating / selecting means 7 integrates event data from the event data recorded on the recording medium of the first information storage / reproducing unit 5 such that a plurality of events become one event group. Or
Event data relating to the first event is selected and copied and recorded in the second information storage / reproduction unit 8.

Here, the type of event or event group, the time of occurrence of the event, the end time of the event, the second information storage / reproduction unit 8
Is created as the second index data, and stored in the recording medium of the second information storage / reproduction unit 8. Thus, the first information storage / reproduction unit 5 and the second
Using the information accumulation / reproduction unit 8, accumulation is performed sequentially in a hierarchical manner.

Next, a method of determining event data to be integrated or selected from event data recorded on the recording medium of the first information storage / reproduction unit 5 will be described. FIG. 3 is an explanatory diagram illustrating a first method of determining an event to be integrated / selected.
In the first determination method, the event integration / selection condition to be set is a time interval of the same event that occurs twice consecutively, and the event integration / selection means 7 continuously generates the event at an interval shorter than this time interval. A plurality of the same events are integrated as one event group. In this case, the selection conditions are similar to those of a human memory mechanism that frequently memorizes frequently occurring events.

For example, as shown in FIG.
The time intervals at which the events with the event IDs 2 and 3 stored in the recording medium of the information storage / reproduction unit 5 occur, and the time intervals at which the events with the event IDs 3 and 4 occur, respectively. It occurs continuously at intervals shorter than the time interval DT. In this case, event data relating to the events whose event IDs are 3 and 4 are selected and further integrated as one event group P. Such a time interval can be calculated using the occurrence time and the end time of the first index data. For example, an intermediate time between the occurrence time and the end time in each event is obtained, and a difference between these intermediate times is set as a time interval. Various other calculations can be made.

The event having the event ID of 2 is copied to the second information storage / reproduction unit 8 as a single event A. Since the event having the event ID of 1 (event B) is not included in the event integration / selection condition, the event information is not copied to the second information storage / reproduction unit 8. After the completion of these processes, of the event data on the recording medium of the first information storage / reproduction unit 5, event data that has not been integrated / selected is deleted as unnecessary.

Next, another method of determining event data to be integrated or selected from time-series continuous event data will be described. FIG. 4 is an explanatory diagram illustrating a second method of determining an event to be integrated / selected. In the second determination method, the event integration / selection condition to be set is a period in a case where normal events occur continuously without any event to be monitored. Means 7
When the event occurs beyond this period, the events in the normal time are collectively integrated as one event group. in this case,
Event data for a normal event may be stored as definition data, and even if event data is not defined, all events that occur continuously for a predetermined period may be collectively defined as normal events. . In this case, the selection condition is similar to that of a human memory mechanism that frequently remembers what frequently occurs (so-called nothing happens).

When the predetermined period is set to L, for example, FIG.
As shown by, no event occurs from the event data stored in the recording medium of the first information storage / reproduction unit 5.
That is, event data in a state where normal events occur continuously beyond the period L are integrated as one event group Q. If there is a section that satisfies this event condition using the first index data of the recording medium of the first information storage / reproduction unit 5, this section is registered in the second information storage / reproduction unit 8 as an event group Q.

However, since the event data for this event group is the same data, in order to increase the recording capacity,
For example, in the case of data relating to an image, event data for at least one frame is used, or event data is not recorded at all. Since the event data during normal times is event data when no event to be monitored occurs, even if only index data is used, the event data sufficiently performs a monitoring function when an event is retrieved later. These event groups Q
The event relating to is basically copied to the second information storage / reproduction unit 8 as the second index data. After the completion of these processes, the event information is deleted from the first information storage / playback unit 5 as unnecessary for event data that has not been integrated / selected.

Next, another method for determining event data to be integrated or selected from time-series continuous event data will be described. FIG. 5 is an explanatory diagram illustrating a third method of determining an event to be integrated / selected. In the third determination method, the event integration / selection condition to be set is the number of times that the same plural events repeatedly occur within a predetermined time interval or less. When the number of occurrences is large, a plurality of the same events are collectively integrated as one event group. In this case, the selection conditions are similar to those of a human memory mechanism that frequently memorizes frequently occurring events.

For example, assuming that the number of times to be set is M, as shown in FIG. 5, out of the event data stored in the recording medium of the first information storage / reproduction unit 5, event A
These events A
Are integrated as an event group R. First information storage / reproduction unit 5
Inspection is performed using the first index data of the recording medium, and if there is a section that satisfies the event condition, this section is registered in the second information storage / reproduction unit 8 as an event group R. The event data related to the event group R is stored in the second information storage / reproduction unit 8.
Is copied to After these processes are completed, the first information storage and
Event information that has not been integrated / selected is deleted from the reproducing unit 5 as unnecessary.

Next, another method of determining event data to be integrated or selected from time-series continuous event data will be described. FIG. 6 is an explanatory diagram illustrating a fourth method of determining an event to be integrated / selected. In this fourth decision method,
The event integration / selection condition to be set is the number of times when a plurality of different events occur alternately within a predetermined time interval or less. The event integration / selection means 7 sets a plurality of identical events when the number of occurrences of the event is larger than this number. Events are collectively integrated as one event group. In this case, the selection condition is a selection condition similar to a human memory mechanism that well remembers frequent occurrences that occur repeatedly and alternately.

For example, assuming that the number of times is M, as shown in FIG. 6, from the event data stored in the recording medium of the first information storage / reproduction unit 5, an event A and an event B are separated by a time interval DT.
If it is determined that it has occurred M times or more consecutively below,
The plurality of events A and the plurality of events B are integrated as an event group S. The first recording medium of the first information storage / reproduction unit 5
If there is a section that satisfies the event integration / selection condition using the index data, the event data relating to the event group S is copied to the second information storage / reproduction unit 8. After the completion of these processes, the event information is deleted from the first information storage / playback unit 5 as unnecessary for event data that has not been integrated / selected.

Next, a description will be given of a fifth determining method which is another determining method of event data to be integrated or selected from time-series continuous event data. In this fifth decision method,
The event occurrence patterns as described in the third and fourth aspects are also stored in a memory (not shown) of the event integration / selection condition setting means 6, and the event integration / selection means 7 performs the processing when the event occurrence pattern matches the event occurrence pattern. Events are collectively integrated as one event group.

For example, a specific pattern, such as an event A, an event B, an event C, an event B, or an event A, which is difficult to define from the event data stored in the recording medium of the first information storage / reproduction unit 5 And when such a specific pattern occurs, they are integrated as one event group. Such a specific pattern can be arbitrarily selected, and if it is stored in a memory (not shown) of the event integration / selection condition setting means 6, a desired specific pattern can be extracted. In this case, depending on the setting of the pattern, a selection condition similar to a human memory mechanism that frequently remembers frequent occurrences, or a human who clearly memorizes things that rarely occur. Selection conditions similar to the above storage mechanism.

The first information storage / reproducing unit 5
When a specific pattern matching the event integration / selection condition occurs using the index data, the events included in the specific pattern are set as one event group to the second information storage / reproduction unit 8, and the event data relating to these event groups Is copied to the second information storage / reproduction unit 8. After these processes,
From the first information storage / reproduction unit 5, event information that is not an object of integration / selection is deleted as unnecessary.

Next, another method of determining event data to be integrated or selected from time-series continuous event data will be described. FIG. 7 is an explanatory diagram illustrating a sixth method of determining an event to be integrated / selected. In the sixth determination method, the event integration / selection condition to be set is a predetermined cumulative time obtained by accumulating the times at which the same event occurs a plurality of times within a predetermined period. It is characterized in that an event that occurs over a predetermined cumulative time or more or an event that occurs during a time less than the predetermined cumulative time is selected. In this case, if the time is equal to or longer than the predetermined cumulative time, a selection condition similar to a human memory mechanism that frequently remembers what frequently occurs is stored, and it is clearly stored that the event rarely occurs if the cumulative time is less than the predetermined cumulative time. This is a selection condition similar to a human memory mechanism.

For example, if the selection conditions are similar to those of a human memory mechanism that frequently remembers frequent occurrences,
As shown in FIG. 7, from the event data relating to the event A, the event B, and the event C stored in the recording medium of the first information storage / reproduction unit 5, the event whose accumulated time is equal to or longer than the predetermined accumulated time T1 is selected. Select all. If the cumulative time when each event occurs is obtained using the first index data of the first information storage / reproducing unit 5, the cumulative time of the event A and the event C exceeds the predetermined cumulative time T1. Second index data of all event data relating to A and event C is created, and the second index data of these event data is recorded in the second information storage / reproduction unit 8. After the completion of these processes, the event information is deleted from the first information storage / playback unit 5 as unnecessary for event data that has not been integrated / selected. If a selection condition similar to a human memory mechanism that clearly memorizes things that rarely occur is used, an event that occurs in the cumulative time within a predetermined cumulative time T1 or less is event B, and this event B will be stored.

Next, another method of determining event data to be integrated or selected from time-series continuous event data will be described. FIG. 8 is an explanatory diagram illustrating a seventh method of determining an event to be integrated / selected. In the seventh determination method, the event integration / selection condition to be set is a predetermined number of occurrences of a certain event within a predetermined period, and the event integration / selection means 7 sets the number of times equal to or more than the predetermined number of occurrences from a plurality of events. Or an event that occurs less than a predetermined number of times. In this case, if the event occurs more than the predetermined number of occurrences, it becomes a selection condition similar to a human memory mechanism that frequently memorizes the frequent occurrence, and if the event occurs less than the predetermined number of occurrences, it is rare. This is a selection condition similar to a human memory mechanism that clearly remembers what does not occur.

For example, if the selection conditions are similar to those of a human memory mechanism that frequently remembers frequent occurrences,
As shown in FIG. 8, from among the event data related to the event A, the event B, and the event C stored in the recording medium of the first information storage / reproduction unit 5, a predetermined number of occurrences (two times) or more within a certain period is performed. Select the event that occurs by the number of times. If the number of occurrences of each event is obtained using the index data of the first information storage / reproduction unit 5, the number of occurrences of the event A and the event C is two or more, and the event A and the event C are selected. . The second index data of all event data relating to the event A and the event C is created, and the second index data of these event data is recorded in the second information storage / reproduction unit 8.
After the completion of these processes, the event information is deleted from the first information storage / playback unit 5 as unnecessary for event data that has not been integrated / selected. If a selection condition similar to a human memory mechanism that clearly memorizes things that rarely occur is used, an event that occurs less than a predetermined number of times (two times) is an event B, This event B will be stored.

Next, another method of determining event data to be integrated or selected from time-series continuous event data will be described. FIG. 9 is an explanatory diagram for explaining an eighth method for determining an event to be integrated / selected. In this eighth decision method,
The event integration / selection condition to be set is a predetermined occurrence frequency of a certain event within a predetermined period, and the event integration / selection means 7 selects an event occurring at a frequency higher than the predetermined occurrence frequency from a plurality of events. It is. In this case, if the event occurs at a frequency higher than the predetermined occurrence frequency, it becomes a selection condition similar to a human memory mechanism that frequently remembers the frequent occurrence, and if the event occurs at a frequency lower than the predetermined occurrence frequency, it is rare. This is a selection condition similar to a human memory mechanism that clearly remembers what does not occur.

For example, if selection conditions similar to those of a human memory mechanism that clearly memorize things that rarely occur are used, as shown in FIG. From the event data on event A, event B, and event C stored in the medium, an event whose event occurrence frequency within a certain period is equal to or less than a predetermined value is selected. Here, the occurrence frequency is a value obtained by dividing the number of occurrences by a certain period.

When the number of occurrences of each event is obtained by using the first index data of the first information storage / reproduction unit 5, the occurrence frequency of the event B is smaller than the predetermined occurrence frequency (P _th ). Is selected. The second index data of all event data relating to this event B is created, and the second index data of these event data is stored in the second information storage /
It is recorded on the reproduction unit 8. After the completion of these processes, the event information is deleted from the first information storage / playback unit 5 as unnecessary for event data that has not been integrated / selected. If a selection condition similar to a human memory mechanism that frequently stores frequently occurring events is used, the frequency of occurrence of events A and C is higher than a predetermined frequency of occurrence (P _th ). A and event C will be stored.

Next, another method of determining event data to be integrated or selected from time-series continuous event data will be described. FIG. 10 is an explanatory diagram illustrating a ninth method of determining an event to be integrated / selected. In the ninth deciding method, the event integration / selection condition to be set is a predetermined average occurrence period within a predetermined period of a certain event, and the event integration / selection means 7 is longer than the predetermined average occurrence period from a plurality of events. An event occurring in a period or an event occurring in a period shorter than a predetermined average period is selected. in this case,
If the event occurs in a period longer than the predetermined average occurrence period, it becomes a selection condition similar to a human memory mechanism that frequently remembers that it occurs frequently, and if the event occurs in a period shorter than the predetermined average occurrence period, it rarely occurs. This is a selection condition similar to a human memory mechanism that clearly remembers things that do not exist.

For example, if the selection conditions are similar to those of a human memory mechanism that frequently remembers frequent occurrences,
As shown in FIG. 10, from the event data relating to the event A, the event B, and the event C stored in the recording medium of the first information storage / reproduction unit 5, the average occurrence period of a certain event in the monitoring section is determined. Select events longer than the mean occurrence period. Here, the average occurrence period is a value obtained by dividing the total period during which an event occurs by the number of occurrences.

When the number of occurrences of each event is obtained using the first index data of the first information storage / reproduction unit 5, the average occurrence period La of the event A and the average occurrence period Lb of the event B become longer than the predetermined occurrence period L. Event A and Event B are selected. The second index data of all event data relating to the event A and the event B is created, and the second index data of these event data is recorded in the second information storage / reproduction unit 8. After these processes,
From the first information storage / reproduction unit 5, event information that is not an object of integration / selection is deleted as unnecessary. If the selection condition is similar to a human memory mechanism that clearly memorizes things that rarely occur, the average occurrence period Lc of the event C is shorter than the predetermined average occurrence period L, These events C will be stored.

As described above, in the time-series information storage / reproduction device, the integration / selection of events when event data of the first information storage / reproduction unit 5 is moved to the second information storage / reproduction unit 8 has been described. The event integration / selection conditions described in the first to ninth determination methods can be variously combined. For example, only the first event integration / selection condition is used, or all the first to ninth event integration / selection conditions are used. It goes without saying that the event integration / selection conditions by other combinations may be used.

The time-series information accumulating / reproducing apparatus may be configured so that event data can be further integrated and selected and reproduced at the time of reproducing. FIG. 11 is a configuration diagram of a time-series information accumulation / reproduction device that integrates, selects, and reproduces event data. As shown in FIG. 11, in addition to the time-series information accumulating / reproducing apparatus described so far, at least a reproducing event condition setting means 13, an event data selecting means 14, and an event data reproducing means 15 are provided. The reproduction event condition setting means 13 specifies event integration / selection conditions for an event group or an event to be reproduced. The event data selection unit 14 selects event data relating to an event group or an event corresponding to the event integration / selection condition. The event data reproducing means 15 stores the selected event data in the first information storage / reproduction unit 5.
And the second information storage / reproduction unit 8 reproduces the information.

Next, the reproduction event condition will be described. FIG. 12 is an explanatory diagram for explaining a reproduction event condition. It is assumed that the time-series information accumulation / reproduction device has performed event integration / selection at the past time T1, and at the current reproduction start time, as shown in FIG. The event data (events of IDs 11, 12, 13, and 14) during the elapsed period (from time T1 to the reproduction start time) is recorded as it is on the recording medium of the first information storage / reproduction unit 5, and FIG. As shown in b), it is recorded from time T0, which is the monitoring start time, to time T1,
Event data (ID for events that have already been integrated and selected
The events (1, 2) are recorded on the recording medium of the second information storage / reproduction unit 8.

The event data in this case is retrieved by the event data selecting means 14 according to the reproduction event conditions set in the reproduction event condition setting means 13. From these, if the setting is made such that the event data of all events newer than the predetermined time T1 and only the event data of the event group P among the events or the event groups before that are selected as the reproduction event conditions, As shown in FIG. 12C, the event data reproducing unit 15 controls the first information storage / reproduction unit 5 and the second information storage / reproduction unit 8, and the second information storage / reproduction unit 8 Event data relating to the events of IDs 11 to 14 is reproduced from the group P and the first information storage / reproduction unit 5.

Here, it is also possible to adopt a method of reproducing and browsing only the index data without reproducing the event data as the first stage, and reproducing the event data only for an event selected from the index data. In the time-series information accumulating / reproducing apparatus configured as described above, the event integration / selection conditions for the event or event group to be reproduced include all event data stored by the first information accumulating / reproducing unit 5, and By using part or all of the event data of the integrated / selected event data stored by the second information storage / reproduction unit 8, all new events are reproduced and old events are collectively reproduced. In addition, the reproduction can be performed so as to increase the monitoring efficiency.

In the time-series information storage / reproduction device described above, the first information storage / reproduction unit 5 and the second information storage / reproduction unit 8 have two levels of hierarchy. (1) The information storage / reproduction unit 5 includes an upper layer,
The second information storage / reproduction unit 8 forms a lower hierarchy. Further, in order to make these time-series information storage / reproduction devices have three or more levels, a new event integration / selection condition setting means, event integration / By cascading one or more of the selection means, the K-th information storage / reproduction unit, and the K-th storage capacity monitoring means, the number of layers of event storage can be increased using a plurality of stages of information storage / reproduction units. .

FIG. 13 is a configuration diagram when the time-series information storage / reproduction device is cascaded. As shown in FIG. In this configuration, a combination of an event integration / selection event condition 16, an event integration / selection unit 17, a K-th information storage / reproduction unit 18, and a K-th storage capacity monitoring unit 19 is connected in K stages in a cascade. New and detailed event information is stored in the first information storage / reproduction unit 5, and the second information storage / reproduction unit 8 is integrated as event information although the time of occurrence is earlier than that of the first information storage / reproduction unit 5. Event information in a selected and unified form is stored in the K-th information storage / reproduction unit 16 in the past, which is distant in time but has a high degree of importance. The events stored in
The second,..., K-th plurality of layers are combined, and hierarchical storage / reproduction is enabled as a whole.

As described above, in this embodiment, after event data is recorded on a recording medium having a fixed capacity, rearrangement recording is performed so as to reconstruct hierarchical time-series events, and event occurrence over the entire monitoring time is performed. It is possible to provide a time-series information storage / reproduction device that can easily grasp the outline of the situation. Further, it is possible to provide a time-series information accumulation / reproduction device for selecting and reproducing a target event or event group at the time of reproduction. Also,
A time-series information storage / reproducing apparatus that stores information hierarchically and facilitates subsequent searches can be provided. Note that a plurality of information storage / reproduction units include, for example, a case where a single device has a plurality of information storage / reproduction units, such as a case where a plurality of hard disks are provided in one cylinder. I have.

Next, a second embodiment according to the second aspect of the present invention will be described. FIG. 14 is a configuration diagram of the time-series information storage / reproduction device of the present embodiment. As shown in FIG. 14, the time-series information storage / reproduction device includes an event data input unit 21, a data storage unit 22, an event detection unit 23, a writing unit 24, and an information storage / reproduction unit 25.
, An event occurrence frequency calculating unit 26, an event occurrence frequency updating unit 27, an importance calculating unit 28, and an event data sorting unit 29. Further, the writing means 24 includes an event data writing control means 24a and an index data creation means 24b. Further, an event data deleting unit 30 for deleting event data in a recording medium at a time is provided.

The event data input means 21 receives event data relating to events occurring in the monitoring area continuously in a time-series manner. The data storage unit 22 stores definition data of an event to be monitored and occurrence frequency data of the event, and the event detection unit 23 can refer to the definition data at any time. The event detection unit 23 compares the event data output from the event data input unit 21 with the definition data stored in the data storage unit 22 and generates and terminates the events to be monitored from the events occurring in time series. Is detected for a plurality of types of events. The writing unit 24 creates index data including an address related to the type, occurrence time, end time, and recording position of the event so as to refer to the event data later when the event detection unit 23 detects the occurrence of the event. Then, the information storage / reproduction unit 25 is controlled to write both the event data and the index data.

The information storage / reproduction unit 25 records and stores event data and index data on a recording medium under the control of the writing unit 24. After accumulation, desired event data can be reproduced from the recording medium with reference to the index data. This recording medium has a ring buffer structure,
The event data is recorded in the ring buffer, and the index data is recorded in the index data storage. After accumulation, desired event data can be reproduced with reference to the index data. Event occurrence frequency calculation means 2
6 calculates the occurrence frequency of a certain event within a certain period using the index data accumulated in the information accumulation / reproduction unit 25 to generate occurrence frequency data. Event occurrence frequency updating means 2
7 updates the occurrence frequency data generated by the event occurrence frequency calculation means 26 by overwriting the occurrence frequency data in the data storage unit.

The importance calculating means 28 reads out the occurrence frequency data from the data storage unit 22 and calculates the importance of each event in the period. Event data sorting means 2
9 sorts event data on the recording medium of the information storage / reproduction unit 25 in accordance with the level of importance calculated by the importance calculation means 28, and records event data of the same importance continuously on the recording medium. Record the relocation in the area. The event data deletion means 30 is a means for issuing an instruction when deleting event data, and is a push button or the like.

[0097] The time-series information storage /
In the reproducing apparatus, the importance calculating means 28 calculates the event data of the event having a low frequency of occurrence as having a high importance and the event data of the event having a high frequency of occurrence as having a low degree of importance. Is to overwrite and update newly input event data in a continuous area on the event data of low importance. In this case, it is similar to a human memory mechanism that clearly memorizes things that rarely occur, and sets event data of an event with a low frequency of occurrence to be high in importance.

Next, the operation of this embodiment will be described. As shown in FIG. 14, the event data input means 21 includes:
Event data relating to an event occurring in the monitoring area is continuously input in a time series. The event data input means 21 outputs event data to the event detection means 23 and the writing means 24, respectively.

The data storage unit 22 connected to the event detection means 23 stores definition data of an event to be monitored and occurrence frequency data of this event in association with each other. You can refer to the definition data. The event detection unit 23 compares the transmitted event data with the definition data stored in the data storage unit 22 and detects the occurrence of an event to be monitored.

When the event detection unit 23 detects that an event to be monitored has occurred, the event detection unit 23 outputs an event detection result signal to the writing unit 24. The index data creating means 24b of the writing means 24 creates index data for referring to this event data later,
The event data writing control unit 24a controls the information storage / reproduction unit 25 to write both the event data and the index data. This index data is composed of addresses relating to the type, occurrence time, end time, and recording position of each event, similarly to the first index data of the first invention.

Under the control of the writing means 24, the information storage / reproduction unit 25 records and stores the input event data and index data on a recording medium. This information storage / reproduction unit 25
After the storage, the desired event data can be reproduced from the recording medium by referring to the index data.

The time management means (not shown) determines that the recordable capacity has been reduced by the storage capacity monitoring means (not shown) every time a predetermined period has elapsed, or the event data deletion means 30 instructs the deletion of unnecessary event data. Then, the event occurrence frequency calculation means 26 calculates the occurrence frequency of a certain event within a certain time using the index data accumulated in the information accumulation / reproduction unit 25, and generates occurrence frequency data. The event occurrence frequency updating unit 27 updates the occurrence frequency data generated by the event occurrence frequency calculating unit 26 by overwriting the occurrence frequency data in the data storage unit 22.

The importance calculating means 28 is provided in the data storage 22
The importance of each event in that period is calculated using the occurrence frequency data stored in the. With reference to the importance calculated by the importance calculating means 28, the event data sorting means 29 sorts the event data according to the level of the importance, and sorts the event data of the information storage / reproduction unit 25, and the importance is determined. The same event data is relocated and recorded in a continuous area on the recording medium.

With such a time-series information accumulating / reproducing apparatus, when it is determined that the recordable area of the recording medium is small, the event data relating to the infrequently occurring event is preferentially left as having high importance. The event having a high frequency of occurrence is defined as the event data having the lowest importance, and the continuous area on the event data having the low importance is overwritten by newly input event data and updated.

Next, the procedure for storing information will be described in detail. FIG. 15 is an explanatory diagram illustrating a method of creating index data related to event data according to the present embodiment. In this description, it is assumed that image data is used as an example of the event data. In order to simplify the explanation, the image data along the lapse of time shown in FIG. The image data is shown as a line. The method of creating index data when recording compressed image data is the same, and a description thereof will be omitted.

The arrow on the lower side of the image data is the event detecting means 2
3 represents the occurrence time and the end time of the detected event. In FIG. 15A, events A, B, and C are detected as occurrence times T1s, T2s, and T3s, and end times T1e, T2e, and T3e, respectively. FIG.
FIG. 5B is an extract of only the image data at the time of occurrence of an event from the image data of FIG. At this time, the image data in the time zone in which no event occurs is discarded without performing the recording operation on the information storage / reproduction unit 25.

As shown in FIG. 15A, as the image data to be recorded when an event occurs, in addition to recording from the event occurrence time to the end time, in addition to the event data from the event occurrence time T1s before Δt as in event A, Δt of event end time T1e
It is also possible to set so as to save the image including before and after the occurrence of the event, such as until later.

FIG. 15C shows the event information storage / reproduction unit 25.
2 shows a storage example of information stored in the storage device. The details of the index data are: event type, event occurrence time, end time,
The pointer of the image data (the recording start position of the image data and the recording end position of the image data) is stored. The image data is image data at the time of occurrence of an event, and is linked by a pointer of index data. At the time of reproduction, the event information accumulation / reproduction unit 25 can access desired event data by referring to the index data.

As described above, the event information accumulation / reproduction unit 25
Only event data when an event occurs is stored. A procedure for preferentially selecting event data having higher importance from the event data will be described. First,
The calculation procedure by the event occurrence frequency calculation means 26 will be described. FIG. 16 is an explanatory diagram illustrating an example of calculating the event occurrence frequency.

A default frequency of occurrence is set in advance for each event. However, since the frequency of occurrence of an event changes with time, it is necessary to update the frequency of occurrence every fixed time. FIG. 16A shows the passage of time and the event occurrence / end times.
In FIG. 16A, the right side indicates the past, and the left end indicates the present. FIG. 16B shows a monitoring period window for calculating the event occurrence frequency. The occurrence frequency of each event in the monitoring period window is updated while periodically sliding the monitoring period window to the left for a certain period.

FIG. 16C is a table showing changes in the frequency of occurrence of events. The occurrence frequency of an event is determined by dividing the occurrence period of each event in the monitoring period window by the length of the window. In this way, the occurrence frequency of an event as shown in FIG. 16C is updated by overwriting the occurrence frequency data of each event stored in the data storage unit 22 as occurrence frequency data.

The importance calculating means 28 performs an event importance calculating operation using the occurrence frequency data. FIG. 17 is an explanatory diagram illustrating an example of calculating the importance of an event. The importance is calculated by a frequency term I (P) based on the event occurrence frequency P and a time term D based on the elapsed time T from the event occurrence / end to the present.
The weighted average of (T) can be defined as follows.

[0113]

[Formula 1] J (P, T) = α × I (P) + (1−α) × D (T)

However, 0 ≦ α ≦ 1, and both I (P) and D (T) are normalized, and 0 ≦ I (P) ≦
It is assumed that 1,0 ≦ D (T) ≦ 1.

Regarding I (P), it is defined that an event having a lower occurrence frequency has a higher importance. For example, the frequency importance I (P) of the next event whose function occurrence frequency is P is calculated.

[0116]

## EQU2 ## I (P) =-K ₁ log ₂ (K ₂ P)

The above equation corresponds to the information amount when the result generated at the probability P in information theory is known. Frequency of occurrence P = 0.5
In this case, -log ₂ (P) is 1, and as the occurrence frequency P approaches 0, -log ₂ (P) becomes a large value. For an event with the occurrence frequency P = 0, -log ₂ (P) is infinite. It will be great. In this case, using K ₁ and K ₂ , 0 ≦ I (P) ≦ 1
Needs to be normalized to satisfy

For D (T), assuming that the elapsed time from the end of the event P to the present is T, the elapsed time T
With regard to, it can be defined as a continuous function in which the importance is higher as the elapsed time of the event that occurs is shorter and the importance is lower as the elapsed time is longer, for example, as follows.

[0119]

## EQU3 ## D (T) = exp (-kT)

The function D (T) is 1 at T = 0, and monotonically decreases as T increases. T is normalized by a constant, such as the length of the monitoring window,
The elapsed time T is used by multiplying it by a constant k.

FIG. 18 is an explanatory diagram for explaining an example of the importance calculation function. FIG. 18A shows an example of a function for obtaining I (P), and FIG. 18B shows an example of a function for obtaining D (T).
According to J (P, T), the frequency of occurrence is small (I
(P) is large) and the elapsed time is short (D
As (T) becomes smaller), the value of J becomes larger. That is, the degree of importance increases.

As described above, the event importance calculating means 28
For each event, the importance is calculated using the updated occurrence frequency and the elapsed time from the event to the present, and the importance data is output to the event data sorting means 29. The event data sorting means 29 sorts the events recorded on the recording medium of the information storage / reproduction unit 25 in order according to the degree of importance.

As a result of the sorting, event data having the same importance is recorded in a continuous area. Events having a certain importance or less are deleted from the information storage / reproduction unit 25, and newly generated event data is stored. Area to be allocated is secured in a continuous area. It should be noted that two types of thresholds are set for the importance and are deleted from the information storage / reproduction unit 25,
For an event having a lower importance than the first threshold, the information data is deleted from the information storage / reproducing unit 25 while leaving only the index data. Data can also be deleted. Also, a threshold value may be provided for the storage capacity, and event data may be deleted from the one with the lower importance so that a certain capacity can be secured. These configurations are set as appropriate.

According to such a method, it is possible to always save event data of high importance in a fixed storage area for an event that has occurred. FIG. 18 is an explanatory diagram illustrating an example of calculating event importance. In FIG. 18, four detection events A
The importance was calculated for 1, B1, B2, and C2. As a result, although the elapsed time is large, the importance is high in the order of A1 with a low occurrence frequency, C2 with a small occurrence time, B2 with a high occurrence frequency, and B1 with the same occurrence frequency and B1 with an elapsed time greater than B2. In order to secure a new storage area, the database should be deleted in the order of B1, B2, C2, and A1.

By the means described above, all events in the near past remain on the storage medium having a fixed capacity, frequently occurring events are deleted in the medium past, and events with a low frequency of occurrence in the distant past are important. Event / image records summarized in human memory can be recorded in such a way that they are recorded as high degree.

As described above, according to the second embodiment of the second invention, voice data, image data or sensor data at the time of occurrence of an event remains on a storage medium of a fixed capacity in a manner similar to a human storage mechanism. In addition, since it can be searched as a database, efficient monitoring can be performed. In addition, even at the time of takeover of the surveillance staff, it is possible to take over including the image recording of the monitoring result of the previous surveillance staff,
Continuous monitoring becomes possible. If the importance is high in a similar manner to a human memory mechanism that frequently memorizes frequent occurrences, it is possible to respond by changing Expression 1.

The first and second embodiments have been described above. The event data input unit 1 described in the first and second embodiments can adopt a configuration common to the first and second embodiments. This point will be described collectively. First invention and second invention
A more specific configuration of the event data input means 1 (21) which can be commonly used in the present invention will be described. Here, the event data input means 1 (21) can select two configurations. 19 and 20 are configuration diagrams showing specific examples of the event data input means.

As a first specific example, as shown in FIG. 19, an event data input means 1 (21) is an image pickup means 1 for picking up an event in a monitoring area and outputting an analog image signal.
a, sound collecting means 1b which collects sound from an event occurring in the monitoring area and outputs an analog sound signal such as sound and noise, and outputs a sensor signal by directly detecting an event such as a detection sensor / detection switch. Sensing means 1c, and an image signal,
A / D converter 1d that converts an audio signal or a sensor signal from analog to digital to output image data, audio data, or sensor data, and stores image data output from the A / D converter 1d. Image memory 1
e, an audio memory 1f for storing audio data output from the A / D converter 1d, and a detection memory 1g for storing sensor data output from the A / D converter 1d. Image data as event data from the image memory 1e, the audio memory 1f, and the detection memory 1g.
Voice data, sensor data, or data obtained by combining these are output. The event data input means 1 according to the first configuration is configured as described above.

As a second specific example, the imaging means 1a, the sound collecting means 1b, the sensing means 1c, and the A / D conversion means 1d of the first configuration shown in FIG.
h. In this case, an imaging unit 1a, a sound collection unit 1b, a sensing unit 1c, an A / D conversion unit 1d, and a transmission unit 1i are arranged in a monitoring area at a remote place. , Sound data, and sensor data, the transmitting means 1h transmits the event data, and the receiving means 1h receives the transmitted event data. The receiving unit 1h receives event data regardless of whether it is wired or wireless, and outputs image data to the image memory 1e, audio data to the audio memory 1f, and sensor data to the detection memory 1g. The event data input means 1 of the second configuration is configured as described above. Thus, the first,
The present invention can be implemented as the event data input unit 1 (21) having any of the second configurations.

Further, a data compression means (not shown) may be provided between the event data input means 1 (21) and the writing means 4 (24) to increase the recording capacity. In the data compression means, the image data sent from the image memory 1e, the audio data sent from the audio memory 1f, and the detection memory 1
g, always performs data compression processing on the sensor data sent from the
Alternatively, the sensor data may be sent to the event data writing control unit 4a, and the compressed image data, audio data, and sensor data may be stored. In the case where the data compression means is provided, the event data compressed when the first information storage / reproduction unit 5 and the second information storage / reproduction unit 8 (information storage / reproduction unit 25) reproduce the event data. Will be used after being defrosted. Further, the event data may be only image data, only audio data, only sensor data, or a combination of these, as necessary. These data formats are appropriately designed and selected.

The event data input means 1 (21)
The event detection means 3 (23) in the case of using can also adopt the configuration common to the first and second embodiments. This point will be described collectively. First, a first specific example of the event detection unit 3 (23) of the present embodiment will be described. FIG. 21 is a configuration diagram showing a first specific example of the event detection means.

The event detecting means 3 of the first specific example uses an image event detecting means 3a for detecting a change occurring in image data or a distribution of the change. Here, the change occurring in the image data or the distribution of the change is stored in the data storage unit 2 as definition data. In particular,
The image event detection unit 3a shown in FIG. 21 includes a previous frame image memory 3b, a difference unit 3c, a binarization unit 3d, a labeling unit 3e, an area measurement unit 3f, and an area comparison unit 3g. Further, the binarizing means 3d reads out the binarized threshold data from the data storage section 2 and the area comparing means 3g reads out the changed area threshold data from the data storage section 2 as definition data.

The previous frame image memory 3b stores the image data of the previous frame. The difference means 3c calculates difference between the current image data output from the image memory 1e and the image data of one frame before output from the previous frame image memory 3b, and outputs the difference data. The difference data is input to the binarization unit 3d, and based on the binarization threshold set by the binarization threshold data, the binarization unit 3d binarizes the difference data and binarizes the difference data. And output the converted image data.

The labeling means 3e outputs data obtained by further labeling the binarized image data. The area measuring means 3f measures the area using the data and outputs the area value of the change. When the area of the change is larger than the threshold set by the change area threshold data, the area comparison means 3g detects the occurrence of an event and outputs an event occurrence detection signal.

Subsequently, as a second specific example of the event detecting means, an image event detecting means 3a for detecting a motion vector of image data or a distribution of the motion vector is used. FIG.
FIG. 3 is a configuration diagram showing a second specific example of the event detection means. Here, the motion vector of the image data generated in the image data or the distribution of the motion vector is stored in the data storage unit 2 as definition data. The event detecting means of the second specific example shown in FIG. 22 includes a previous frame image memory 3h, a motion detecting means 3i, and a motion comparing means 3j. Further, the motion comparing unit 3j reads the motion pattern distribution model data and the motion detection threshold data from the data storage unit 2 as definition data.

The motion detecting means 3i detects a motion vector from the current image data output from the image memory 1e and the image data of one frame before output from the previous frame image memory 3i. Comparing the motion vector with the motion pattern model data, if a similarity is found above the threshold of the motion detection threshold data,
An event occurrence is detected and an event detection result signal is output. It is assumed that the motion to be detected can be detected based on the direction of the motion in the screen, the magnitude of the motion, the distribution and area of the motion area.

As a third specific example of the event detecting means, an image event detecting means 3a for detecting a specific pattern or pattern distribution in image data is used. FIG.
FIG. 7 is a configuration diagram showing a third specific example of the event detection means. Here, a specific pattern or pattern distribution in the image data is stored in the data storage unit 2 as definition data. As shown in FIG.
a includes an image matching unit 3k and a similarity determination unit 3l. The similarity determination unit 31 reads the template image data and the similarity threshold data from the data storage unit 2 as definition data.

The image matching means 3k includes the image memory 1
The image data output from e is matched, and the matched image data is output to the similarity determination unit 3l. The similarity determination means 31 generates a similarity by comparing the matched image data with the template image data, and when the similarity satisfies the condition by comparing with the similarity threshold value based on the similarity threshold data. , And outputs an event detection result signal as if an event has occurred.

As a fourth specific example of the event detecting means, a voice event detecting means 3m for detecting a change in amplitude or frequency of voice data is used. FIG. 24 shows a fourth example of the event detection means.
FIG. 4 is a configuration diagram showing a specific example of the above. Here, the change in the amplitude and frequency of the audio data is stored in the data storage unit 2 as definition data. As shown in FIG. 24, the voice event detecting means 3m includes an amplitude change detecting means 3n, a frequency change detecting means 3o, and a result integrating means 3p. The amplitude change detecting means 3n reads out the amplitude change threshold data from the data storage unit 2 and the frequency change detecting means 3o reads out the frequency change threshold data from the data storage unit 2 as the definition data.

The sound data output from the sound memory 1f is input to the amplitude change detecting means 3n and the frequency change detecting means 3o. By comparing the amplitude change of the audio data with the amplitude change threshold value, and comparing the frequency change of the audio data with the frequency change threshold value, the amplitude and frequency change degree are input to the result integrating means 3p. . The result integration means 3p integrates these degrees of change and, when it is determined that a change has occurred in the audio data, detects the occurrence of an event and outputs an event detection result signal.

As a fifth specific example of the event detecting means, a voice event detecting means for detecting a pattern such as amplitude and frequency of voice data is used. FIG. 25 is a configuration diagram showing a fifth specific example of the event detection means. Here, patterns such as the amplitude and frequency of audio data are stored in the data storage unit 2 as definition data. As shown in FIG. 25, the voice event detecting unit 3m includes a voice matching unit 3q.
It has. The voice matching means 3q reads out voice pattern data and voice similarity threshold data from the data storage unit 2 as definition data.

The voice data output from the voice memory 1f is output to the voice matching means 3q, and is compared with voice pattern data to determine similarity between amplitude and frequency patterns. If the threshold value is larger than the threshold value, an event detection result signal is output.

As a sixth specific example of the event detecting means, an event direct detecting means 3r such as a sensor for detecting a specific pattern on sensor data which differs depending on the event is used.
FIG. 26 is a configuration diagram showing a sixth specific example of the event detection means. The event direct detection means 3r is provided with an external signal input means 3s.
And comparison means 3t. Here is the detection memory 1g
The specific pattern on the sensor data output from the sensor data and the sensor data output from the external signal input means 3s is stored in the data storage unit 2 as definition data.

The external signal input means 3s uses external devices such as sensors and push buttons. For example, when the push button is turned ON, an ON / OFF signal is output.
In this case, if the sensor data is, for example, data such as 01011, a plurality of events can be detected. The sensor data output from the detection memory 1g and the sensor data output from the external signal input means 3s are:
The signal is output to the comparing means 3t, is compared with the sensor signal pattern data, and when it matches a specific pattern, it is determined that an event has occurred and an event detection result signal is output.

As described above, the event data input means 1 (2
In 1), one of two specific examples can be selected. Further, the event detecting means 3 (23) can select any one of the first to sixth means, and any of the time series information accumulating / reproducing apparatuses combining them can be selected from the first and the second means.
The implementation of the second invention becomes possible.

[0146]

As described above, according to the present invention, storage processing of event data is performed on a recording medium of a fixed capacity so as to increase the search efficiency, and a quick and easy search can be realized later when searching for contents. It is possible to provide an apparatus for storing and reproducing time-series information.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a time-series information storage / reproduction device according to a first embodiment of the present invention.

FIG. 2 is an explanatory diagram illustrating an example of a process of storing time-series event data input to a time-series information storage / playback apparatus according to a first embodiment of the present invention in a first information storage / playback unit;

FIG. 3 is an explanatory diagram illustrating a first method of determining an event to be integrated / selected in the time-series information storage / reproduction device according to the first embodiment of the present invention.

FIG. 4 is an explanatory diagram illustrating a second method of determining an event to be integrated / selected in the time-series information storage / reproduction device according to the first embodiment of the present invention.

FIG. 5 is an explanatory diagram illustrating a third method of determining an event to be integrated / selected in the time-series information storage / playback device according to the first embodiment of this invention.

FIG. 6 is an explanatory diagram illustrating a fourth method of determining an event to be integrated / selected in the time-series information storage / reproduction device according to the first embodiment of the present invention.

FIG. 7 is an explanatory diagram illustrating a sixth method of determining an event to be integrated / selected in the time-series information storage / reproduction device according to the first embodiment of the present invention.

FIG. 8 is an explanatory diagram illustrating a seventh method of determining an event to be integrated / selected in the time-series information storage / playback apparatus according to the first embodiment of this invention.

FIG. 9 is an explanatory diagram illustrating an eighth method for determining an event to be integrated / selected in the time-series information storage / playback apparatus according to the first embodiment of this invention.

FIG. 10 is an explanatory diagram illustrating a ninth method of determining an event to be integrated / selected in the time-series information storage / reproduction device according to the first embodiment of this invention.

FIG. 11 is a configuration diagram of a time-series information storage / reproduction device that integrates, selects, and reproduces event data in the time-series information storage / reproduction device of the first embodiment of the present invention.

FIG. 12 is an explanatory diagram illustrating playback event conditions of the time-series information storage / playback device according to the first embodiment of the present invention.

FIG. 13 is a configuration diagram when the time-series information storage / reproduction device according to the first embodiment of the present invention is connected in cascade.

FIG. 14 is a configuration diagram of a time-series information storage / reproduction device according to a second embodiment of the present invention.

FIG. 15 is an explanatory diagram illustrating a method of creating index data related to event data according to a second embodiment of the present invention.

FIG. 16 is an explanatory diagram illustrating an example of calculating an event occurrence frequency according to the second embodiment of the present invention.

FIG. 17 is an explanatory diagram illustrating an example of calculating importance according to the second embodiment of the present invention.

FIG. 18 is an explanatory diagram illustrating an example of calculating importance according to the second embodiment of the present invention.

FIG. 19 is a configuration diagram showing a first specific example of event data input means common to the first and second embodiments of the present invention.

FIG. 20 is a configuration diagram showing a second specific example of the event data input means common to the first and second embodiments of the present invention.

FIG. 21 is a configuration diagram showing a first specific example of event detection means common to the first and second embodiments of the present invention.

FIG. 22 is a configuration diagram showing a second specific example of the event detection means common to the first and second embodiments of the present invention.

FIG. 23 is a configuration diagram showing a third specific example of the event detection means common to the first and second embodiments of the present invention.

FIG. 24 is a configuration diagram showing a fourth specific example of the event detection means common to the first and second embodiments of the present invention.

FIG. 25 is a configuration diagram showing a fifth specific example of the event detection means common to the first and second embodiments of the present invention.

FIG. 26 is a configuration diagram showing a sixth specific example of the event detection means common to the first and second embodiments of the present invention.

FIG. 27 is a configuration diagram showing a conventional time-series information storage / reproduction device.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 Event data input means 1a Imaging means 1b Sound collection means 1c Sensing means 1d A / D conversion means 1e Image memory 1f Audio memory 1g Detection memory 1h Receiving means 1i Transmitting means 2 Data storage part 3 Event detecting means 3a Image event detecting means 3b Previous frame image memory 3c Difference means 3d Binarization means 3e Labeling means 3f Area measurement means 3g Area comparison means 3h Previous frame image memory 3i Motion detection means 3j Motion comparison means 3k Image matching means 31 Similarity determination means 3m Voice event detection means 3n amplitude change detection means 3o frequency change detection means 3p result integration means 3q voice matching means 3r event direct detection means 3s external signal input means 3t comparison means 4 writing means 4a event data writing control means 4b index data creation means 5 first Information storage・ Reproduction unit 6 Event integration / selection condition setting means 7 Event integration / selection means 8 Second information storage / reproduction unit 9 First storage capacity monitoring means 10 Second storage capacity monitoring means 11 Time management means 12 Monitoring start input means 13 Reproduction Event condition setting means 14 Event data selection means 15 Event data reproduction means 16 Event integration / selection condition setting means 17 Event integration / selection means 18 Kth information storage / reproduction unit 19 Kth storage capacity monitoring means 21 Event data input means 22 Data Storage unit 23 Event detection unit 24 Writing unit 24a Event data writing control unit 24b Index data creation unit 25 Information storage / reproduction unit 26 Event occurrence frequency calculation unit 27 Event occurrence frequency update unit 28 Importance calculation unit 29 Event data sorting unit 30 Event data deletion means

 ──────────────────────────────────────────────────続 き Continuing from the front page (72) Inventor Hideaki Uekusa 1Fuji-cho, Hino-shi, Tokyo Inside FFC Co., Ltd. No.1 Fuji Electric Co., Ltd. F-term (reference) 5C053 FA23 FA27 GA11 JA01 JA22 JA24 LA01 5C054 AA01 CE12 DA01 DA07 DA08 EG01 FC01 FC05 FC12 FC15 GA01 GA04 GB01 HA18

Claims (21)

[Claims] 1. Event data input means for inputting event data relating to an event continuously in a time series; a data storage unit for storing event definition data; and event data output from the event data input means. Event detection means for detecting the occurrence and end of an event from among events occurring in time series for a plurality of types of events, by comparing the event data with definition data stored in the data storage unit; When the event data is detected, first index data including an address related to the type, occurrence time, end time, and recording position of each event is created to refer to the event data later. And writing the event data and the first index data on a recording medium under the control of the writing means. Thereafter, a first information storage / reproduction unit that reproduces desired event data from the recording medium with reference to the first index data, and one event group from the event data recorded in the first information storage / reproduction unit Event integration / selection condition setting means for setting event integration / selection conditions for extracting event data of a plurality of events or event data of one event to be integrated as storage targets, and the event integration / selection condition setting means Event data of a plurality of events based on the event integration / selection conditions set by
Integration of the event data of the event group of the above, or the selection of the event data of the one event, and the type, occurrence time, and end time of the event or the event group to refer to these event or the event data of the event group later An event integration / selection unit that creates second index data including an address related to a recording position, and controls to write the event data and the second index data together, and that the event integration / selection unit controls the second index data. 1) Event information and second index data transferred from the information storage / reproduction unit are recorded and accumulated on a recording medium, and after accumulation, second information accumulation / reproduction for reproducing desired event data with reference to the second index data. A time-series information storage / playback device comprising at least: a first information storage / playback unit that stores all event data newer than a certain point in time. Resides, and series information storage and reproducing apparatus when characterized by saving in terms of the integrated and select the old event data from the time that the second information storage and reproduction unit. 2. The time integration / selection device according to claim 1, wherein the event integration / selection condition is a time interval of the same event occurring twice consecutively. Is a time-series event accumulation / reproduction apparatus characterized in that a plurality of identical events continuously occurring at a time interval shorter than this time interval are integrated as one event group. 3. The time series information accumulation / reproduction device according to claim 1, wherein the event integration / selection condition is a period when a normal event occurs continuously, and the event integration / selection is performed. The time-series event storage / reproduction device is characterized in that the means integrates normal events that occur beyond this period into one event group. 4. The time-series information accumulating / reproducing apparatus according to claim 1, wherein the event integration / selection condition is a number of times when the same plurality of events repeatedly occur at a predetermined time interval or less, The time series event accumulation / reproduction device, wherein the event integration / selection means integrates the plurality of identical events into one event group when the number of occurrences of the event is greater than the number of times. 5. The time-series information accumulating / reproducing apparatus according to claim 1, wherein the event integration / selection condition is a number of times when a plurality of different events occur alternately within a predetermined time interval, The time series event accumulation / reproduction device, wherein the event integration / selection means integrates the plurality of identical events into one event group when the number of occurrences of the event is greater than the number of times. 6. The time-series information storage / reproduction device according to claim 1, wherein the event integration / selection condition is whether or not the event integration / selection condition matches a predetermined event occurrence pattern. A time-series event accumulation / reproduction device, wherein the selection means integrates events in a case where the occurrence pattern of the event matches, into one event group; 7. The time-series information accumulating / reproducing apparatus according to claim 1, wherein the event integration / selection condition is a predetermined cumulative time obtained by accumulating a plurality of times during which a same event occurs within a predetermined period. The time series event accumulation / reproduction device is characterized in that the event integration / selection means selects, from a plurality of events, an event that occurs for a time longer than the predetermined cumulative time or an event that occurs for a time less than the predetermined cumulative time. 8. The time integration / selection device according to claim 1, wherein the event integration / selection condition is a predetermined number of occurrences of an event within a predetermined period. A time-series event storage / reproduction device, which selects an event occurring more than the predetermined number of times or an event occurring less than the predetermined number of times from a plurality of events. 9. The time series information storage / reproduction device according to claim 1, wherein the event integration / selection condition is a predetermined occurrence frequency of an event within a predetermined period, and the event integration / selection means A time-series event storage / reproduction device, wherein an event occurring at a frequency higher than the predetermined occurrence frequency or an event occurring at a frequency lower than the predetermined occurrence frequency is selected from a plurality of events. 10. The time series information accumulation / reproduction apparatus according to claim 1, wherein the event integration / selection condition is a predetermined average period of occurrence of an event within a predetermined period, and the event integration / selection is performed. A time-series event accumulation / reproduction device, wherein the means selects an event occurring in a period longer than the predetermined average period or an event occurring in a period shorter than the predetermined average period from a plurality of events. 11. A reproduction event condition setting means for designating a reproduction event condition for an event group or an event to be reproduced in the time-series information storage / reproduction device according to any one of claims 1 to 10. Event data selecting means for selecting an event group or event data relating to the event corresponding to the reproduction event condition; and reproducing the selected event data from the first information storage / reproduction unit and the second information storage / reproduction unit. Event data reproducing means, wherein a reproduction event condition for the event group or event to be reproduced includes all event data stored by the first information storage / reproduction unit, and the second information storage / reproduction Time-series information storage, characterized in that part or all of the event data selected and integrated by the
Playback device. 12. The time-series information storage / reproduction device according to claim 1, wherein an event integration / selection condition setting is newly provided after the second information storage / reproduction unit. Means, event integration / selection means, and information storage / reproduction unit in cascade, and K (K is a natural number)
A time-series information storage / reproduction apparatus characterized in that the number of layers when event data is stored is increased by using an information storage / reproduction unit of a stage. 13. Event data input means for continuously inputting event data relating to an event in a time-series manner; a data storage unit for storing event definition data and occurrence frequency data of the event; Event detection means for comparing the event data output from the means and the definition data stored in the data storage unit and detecting the occurrence and end of the event from among the events occurring in time series for a plurality of types of events, When the occurrence of an event is detected by the event detection means, index data including addresses relating to the type, occurrence time, end time, and recording position of each event is created for later reference to the event data. Writing means for controlling data and index data to be written together; and controlling event data and index data by controlling the writing means. An information storage / reproducing unit that records and accumulates the data on the recording medium and reproduces desired event data from the recording medium by referring to the index data after the accumulation; and an event using the index data accumulated in the information storage / reproducing unit. Event occurrence frequency calculation means for calculating occurrence frequency within a certain period of time and generating occurrence frequency data, and updating the occurrence frequency data generated by the event occurrence frequency calculation means with occurrence frequency data in the data storage unit An event occurrence frequency updating unit, an importance calculating unit that reads the occurrence frequency data from the data storage unit and calculates the importance of each event in the period, and changes the importance calculated by the importance calculating unit. Accordingly, the event data on the recording medium of the information storage / reproducing unit is sorted, and event data having the same importance is re-read to a continuous recording area on the recording medium. A time-series information accumulating / reproducing device comprising at least an event data sorting means for arranging and recording, wherein the information accumulating / reproducing unit has an event of low importance based on the importance calculated by the importance calculating means. A time-series information accumulating / reproducing apparatus characterized in that a continuous area on data is updated by overwriting newly input event data. 14. The time-series information accumulating / reproducing apparatus according to claim 1, wherein said event data input means outputs a sound signal by sound collection, Imaging means for outputting an image signal, or sensing means for outputting a continuously generated sensor signal, and converting these audio signals, image signals, or sensor signals into audio data, image data, or sensor data, respectively. A time-series information storage / reproduction device comprising: 15. The time-series information accumulating / reproducing apparatus according to claim 1, wherein the event data inputting means includes audio data and image data transmitted via a communication facility. Or a time-series information storage / reproduction device, which is a receiving means for receiving and outputting sensor data. 16. The time-series information storage / reproduction device according to claim 14, wherein the event data output by the event data input means includes at least image data of an event, Storing data relating to a change or a change distribution as definition data, wherein the event detecting means includes an image event detecting means for detecting a change or a change distribution occurring in the image data; apparatus. 17. The time-series information storage / reproduction device according to claim 14, wherein the event data output by the event data input means includes at least image data of the event, A specific motion pattern or motion pattern distribution generated as definition data is stored, and the event detection means includes image event detection means for detecting a specific motion pattern or motion pattern distribution generated in image data. Time-series information storage / reproduction device. 18. The time-series information accumulating / reproducing apparatus according to claim 14, wherein the event data output from the event data input means includes at least image data of the event, Storing a specific pattern or pattern distribution in the image data as definition data, wherein the event detecting means includes an image event detecting means for detecting the specific pattern or the pattern distribution in the image data;・ Playback device. 19. The time-series information accumulating / reproducing apparatus according to claim 14, wherein the event data output by the event data input means includes at least voice data of the event, A time-series information accumulating / reproducing apparatus which stores a change in audio data as definition data, and wherein the event detection means includes an audio event detection means for detecting a change in audio data. 20. The time-series information accumulating / reproducing apparatus according to claim 14, wherein the event data output by the event data input means includes at least audio data of the event, A time-series information storage / reproduction device, wherein a specific pattern on audio data is stored as definition data, and the event detection means includes an audio event detection means for detecting a specific pattern on the audio data. 21. The time-series information storage / reproduction device according to claim 14, wherein the event data output from the event data input means includes at least sensor data of an event, A time-series information accumulating / reproducing apparatus, wherein a specific pattern on sensor data is stored as definition data, and the event detecting means includes an event direct detecting means for detecting a specific pattern on sensor data.