CN116887057B - Intelligent video monitoring system - Google Patents

Abstract

The invention provides an intelligent video monitoring system, which comprises: the data acquisition module is used for acquiring monitoring videos of multiple angles of a target monitoring area based on multiple shooting devices, and splicing the monitoring videos to generate a panoramic monitoring video; the feature extraction module is used for marking a monitoring target of the panoramic monitoring video, determining a reference live-action corresponding to the monitoring target, performing action positioning on the monitoring target based on the reference live-action, and extracting the motion feature of the monitoring target according to the motion positioning; and the judging and early warning module is used for judging whether the monitoring target has abnormal behaviors based on the motion characteristics, and if so, generating a corresponding abnormal early warning notice and sending the monitoring management center based on the abnormal behaviors of the monitoring target. The method and the device can generate panoramic monitoring video of the target monitoring area, automatically realize automatic marking of various monitoring targets, can be applied to various monitoring scenes, and finish synchronous monitoring of various monitoring targets in the same monitoring area.

Description

Intelligent video monitoring system

Technical Field

The invention relates to the technical field of video monitoring, in particular to an intelligent video monitoring system.

Background

The monitoring system plays an increasing role in production and life, and becomes an indispensable safety barrier in life of people, and the needs of public areas such as banks, supermarkets, shops, storefronts, factories, schools, communities, internet cafes, public transportation and the like for video monitoring are self-evident. Meanwhile, the video monitoring video is also widely applied to safety monitoring of working areas such as factories, construction sites and the like, and early warning is carried out on dangerous behaviors in time, so that the accident rate is reduced. However, the limitation of the video monitoring angle and the rotation time difference can make the behavior action of the monitored target in the video monitoring range unable to be completely presented, so that the behavior of the monitored target cannot be accurately identified and pre-warned.

Disclosure of Invention

The intelligent video monitoring system provided by the invention can generate panoramic monitoring video of the target monitoring area, ensure that the behavior action of each monitoring target in the target monitoring area can be completely seen, and effectively improve the accuracy of monitoring target behavior identification. According to the monitoring setting information, the monitoring target is automatically marked on the panoramic monitoring video picture, automatic marking of the monitoring target is achieved, the monitoring system can be applied to various monitoring scenes, synchronous monitoring of various monitoring targets in the same monitoring area is completed, the reference real scene corresponding to the monitoring target is determined, pre-judging of abnormal behaviors related to the monitoring target is completed, then the monitoring target is subjected to action positioning based on the reference real scene, the motion characteristics of the monitoring target are extracted according to the motion positioning, the actual motion track of the monitoring target and the motion behaviors relative to the reference real scene can be accurately identified, the accuracy of the motion characteristics of the monitoring target is effectively improved, and therefore accuracy of judging the abnormal behavior identification of an early warning module is provided, and accurate early warning of the abnormal behaviors is achieved.

The invention provides an intelligent video monitoring system, which comprises:

the data acquisition module is used for acquiring monitoring videos of multiple angles of a target monitoring area based on multiple shooting devices, and splicing the monitoring videos to generate a panoramic monitoring video;

the feature extraction module is used for marking a monitoring target on the panoramic monitoring video, determining a reference live-action corresponding to the monitoring target, performing action positioning on the monitoring target based on the reference live-action, and extracting the motion features of the monitoring target according to the motion positioning;

and the judging and early warning module is used for judging whether the monitoring target has abnormal behaviors based on the motion characteristics, and if so, generating a corresponding abnormal early warning notice and sending the monitoring management center based on the abnormal behaviors of the monitoring target.

Preferably, the data acquisition module in an intelligent video monitoring system comprises: the installation recommendation sub-module is used for generating a shooting device installation strategy corresponding to each monitoring position in the target monitoring area according to the actual monitoring requirement of the target monitoring area, and comprises the following steps:

the first determining unit is used for obtaining effective monitoring distances corresponding to all monitoring positions of the target monitoring area and determining angle rotation homing time of the shooting equipment based on the scene preset speed and the effective monitoring distances;

A second determining unit, configured to determine a monitoring angle of each photographing apparatus based on the angular rotation homing time and an angular velocity of the camera;

the strategy generation unit is used for determining the target quantity of shooting equipment corresponding to the target monitoring area according to the panoramic monitoring angle corresponding to the monitoring position and the monitoring angle, generating equipment installation recommendation strategies corresponding to the monitoring position and sending the equipment installation recommendation strategies to the monitoring management center;

and the parameter setting unit is used for automatically setting shooting parameters of all shooting equipment according to the monitoring angles of all the shooting equipment and the actual installation positions of the shooting equipment after the shooting equipment is installed.

Preferably, the data acquisition module in an intelligent video monitoring system further comprises:

the video acquisition unit is used for acquiring monitoring videos of different angles based on a plurality of shooting devices of the target monitoring area;

the sequence determining unit is used for acquiring shooting parameters corresponding to each shooting device in the target monitoring area and determining a splicing sequence of pictures corresponding to the monitoring video based on the shooting parameters;

the video stitching unit is used for stitching pictures of the monitoring video at the same moment according to the stitching sequence to obtain panoramic pictures, and generating panoramic monitoring videos according to the time axis sequence based on the panoramic pictures.

Preferably, in an intelligent video monitoring system, the video stitching unit includes:

the direction determining subunit is used for respectively determining a monitoring angle starting position and a monitoring angle ending position of each monitoring video, determining a monitoring period corresponding to each monitoring video based on the monitoring angle starting position and the monitoring angle ending position, and generating a forward displacement vector and a reverse displacement vector of the shooting equipment;

the forward splicing unit is used for acquiring a first monitoring sub-video corresponding to the upper half monitoring period, acquiring a first video picture of the first monitoring sub-video frame by frame, and carrying out picture splicing on the first video picture based on a forward displacement vector to acquire a forward picture of the monitoring angle;

acquiring all forward pictures of all shooting devices in a target monitoring area in the same monitoring period, and splicing the forward monitoring pictures according to the splicing sequence to obtain a forward panoramic picture;

the reverse splicing unit is used for acquiring a second monitoring sub-video corresponding to the lower half monitoring period, acquiring a second video picture of the second monitoring sub-video frame by frame, and carrying out picture splicing on the second video picture based on a reverse displacement vector to acquire a reverse picture of the monitoring angle;

Acquiring all reverse images of all shooting devices in a target monitoring area in the same monitoring period, and splicing the reverse monitoring images according to the reverse sequence of the splicing sequence to obtain a reverse panoramic image;

and the panoramic display subunit is used for alternately displaying the forward panoramic picture and the reverse panoramic picture based on the time axis sequence to generate a panoramic monitoring video.

Preferably, in an intelligent video monitoring system, the feature extraction module includes:

the target marking unit is used for determining object characteristics corresponding to the video monitoring object based on the monitoring setting information, carrying out target identification on the panoramic monitoring video based on the object characteristics, and carrying out monitoring target marking according to a target identification result;

the reference determining unit is used for acquiring monitoring safety standards based on the monitoring setting information, determining a safety identification reference object, and carrying out reference marking in the panoramic monitoring video based on the safety identification reference object to obtain a reference live-action;

the reference corresponding unit is used for determining reference live-action corresponding to each monitoring target according to the nearby principle;

the action marking unit is used for coding the monitoring targets based on the monitoring target marks to obtain a monitoring target sequence, and according to the monitoring target sequence, action positioning is carried out on different monitoring targets in the panoramic monitoring video successively, and the motion characteristics of different monitoring targets are extracted by combining the relative relation between the monitoring targets and the corresponding reference live-action.

Preferably, in an intelligent video monitoring system, the reference determining unit includes:

the standard acquisition subunit is used for determining a plurality of monitoring contents corresponding to the target monitoring area and standard setting information corresponding to each monitoring content based on the monitoring setting information;

generating corresponding safety standards according to the standard setting information respectively, and acquiring the safety standards corresponding to all monitoring contents to generate monitoring safety standards corresponding to the monitoring positions;

the reference determination subunit is used for determining a plurality of safety identification references corresponding to the monitoring positions according to the monitoring safety standards and acquiring object characteristics corresponding to each safety identification reference;

and the reference calibration unit is used for carrying out reference marking in the panoramic monitoring video based on the object characteristics, and taking the real scene of the reference marking as the reference real scene currently corresponding to the monitoring position.

Preferably, in an intelligent video monitoring system, the action marking unit includes:

the marking subunit is used for marking different monitoring targets in the video frames corresponding to the current panoramic monitoring video according to the target monitoring sequence to obtain a first action, and comparing the first action position with the reference live-action position to obtain a first relative relation;

The acquisition subunit is used for acquiring second actions and second relative relations corresponding to different monitoring targets on a previous frame picture of a video frame corresponding to the current panoramic monitoring video according to the target monitoring sequence;

and the comparison subunit is used for comparing the first action with the second action to obtain action change characteristics, comparing the first relative relation with the second relative relation to obtain position change characteristics, and generating the movement characteristics of the monitoring target based on the action change characteristics and the position change characteristics.

Preferably, in an intelligent video monitoring system, the reference determining unit further includes:

the reference updating subunit is used for updating the reference live-action corresponding to the monitoring target based on the nearby principle after the first relative relation of the video frame corresponding to the current panoramic monitoring video, determining the current live-action of the monitoring target, taking the first action as the second action when the current reference live-action is consistent with the original reference live-action, taking the first relative relation as the second relative relation, and waiting for the call of the acquisition subunit;

otherwise, comparing the current reference live-action position with the first action position, obtaining a second relative relation according to a comparison result, and taking the first action as a second action to wait for the call of the acquisition subunit.

Preferably, the judging and early warning module in an intelligent video monitoring system comprises:

the standard determining unit is used for determining all the corresponding to-be-selected abnormal behaviors of different monitoring targets in the target monitoring area and the abnormal motion characteristics corresponding to the to-be-selected abnormal behaviors according to the monitoring setting information, and determining the corresponding target abnormal behaviors of the different monitoring targets and the corresponding target abnormal motion characteristics based on the reference real scenes corresponding to the different monitoring targets;

the characteristic comparison unit is used for comparing the motion characteristics of different monitoring targets with the corresponding abnormal motion characteristics of the targets to obtain the characteristic similarity;

the judging and early warning unit is used for judging that the monitoring target behavior is abnormal when the feature similarity is larger than or equal to a preset value, and generating a corresponding abnormal early warning notice according to the target abnormal behavior corresponding to the monitoring target to send a monitoring management center;

and when the feature similarity is smaller than a preset value, judging that the monitoring target acts normally, and continuously monitoring the monitoring target.

Preferably, in an intelligent video monitoring system, further comprising: the remote control module is used for monitoring the remote control of the management center and accurately checking the current actual condition of a certain position in the target monitoring area, and comprises the following steps:

The target selecting unit is used for selecting an amplified viewing target in the panoramic monitoring video by the monitoring management center;

the target determining unit is used for determining a panoramic display subarea where the amplified reference target is located, determining a monitoring position and a corresponding monitoring angle based on the panoramic display subarea, and determining target shooting equipment according to the monitoring position and the corresponding monitoring angle;

the focusing and amplifying unit is used for controlling and adjusting the shooting focal length of the target shooting equipment to focus and shoot the amplified viewing target so as to obtain a local amplified video;

and the sending display unit is used for sending the local amplified video to a preset display position for local viewing for display.

Compared with the prior art, the invention has at least the following beneficial effects:

according to the invention, the monitoring videos of multiple angles of the target monitoring area are obtained through multiple shooting devices of the data acquisition module, and the monitoring videos are spliced to generate the panoramic monitoring video, so that the behavior action of each monitoring target in the target monitoring area can be completely seen, and the accuracy of monitoring target behavior identification is effectively improved; the feature extraction module is used for marking the monitoring target on the panoramic monitoring video picture, so that the monitoring equipment can apply various monitoring scenes, synchronous monitoring of various monitoring targets in the same monitoring area is completed, the reference real scene corresponding to the monitoring target is determined, the prejudgment of abnormal behavior related to the monitoring target is completed, then the monitoring target is subjected to action positioning based on the reference real scene, the motion features of the monitoring target are extracted according to the motion positioning, the actual motion track of the monitoring target and the motion behavior relative to the reference real scene can be accurately identified, the accuracy of the operation features of the monitoring target is effectively improved, and the accuracy of the abnormal behavior identification of the judgment early warning module is provided; and judging whether the monitoring target has abnormal behaviors or not through the judging and early warning module based on the motion characteristics, if so, generating a corresponding abnormal early warning notice to send a monitoring management center based on the abnormal behaviors, automatically completing abnormal behavior monitoring in a target monitoring area, reducing the workload of monitoring and checking staff, avoiding errors of manual judgment, and reducing the misjudgment rate of the abnormal behaviors.

Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims thereof as well as the appended drawings.

The technical scheme of the invention is further described in detail through the drawings and the embodiments.

Drawings

The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention. In the drawings:

FIG. 1 is a schematic diagram of an intelligent video monitoring system according to the present invention;

FIG. 2 is a schematic diagram of a data acquisition module of an intelligent video monitoring system according to the present invention;

FIG. 3 is a schematic diagram of a feature extraction module of an intelligent video surveillance system according to the present invention;

fig. 4 is a schematic diagram of a judgment and early warning module of an intelligent video monitoring system according to the present invention.

Detailed Description

The preferred embodiments of the present invention will be described below with reference to the accompanying drawings, it being understood that the preferred embodiments described herein are for illustration and explanation of the present invention only, and are not intended to limit the present invention.

Example 1: the invention provides an intelligent video monitoring system, if shown in fig. 1, comprising:

the data acquisition module is used for acquiring monitoring videos of multiple angles of a target monitoring area based on multiple shooting devices, and splicing the monitoring videos to generate a panoramic monitoring video;

the feature extraction module is used for marking a monitoring target on the panoramic monitoring video, determining a reference live-action corresponding to the monitoring target, performing action positioning on the monitoring target based on the reference live-action, and extracting the motion features of the monitoring target according to the motion positioning;

and the judging and early warning module is used for judging whether the monitoring target has abnormal behaviors based on the motion characteristics, and if so, generating a corresponding abnormal early warning notice and sending the monitoring management center based on the abnormal behaviors of the monitoring target.

In this embodiment, the target monitoring area refers to an area monitored by video.

In this embodiment, the monitoring video refers to a monitoring video within a corresponding monitoring angle range of each photographing device.

In this embodiment, the panoramic monitoring video refers to a panoramic video of a target monitoring area obtained by stitching all the monitoring videos.

In this embodiment, the monitoring target refers to a monitored object corresponding to the target monitoring area.

In this embodiment, the reference live-action refers to a reference actual scene in the panoramic surveillance video as a judgment of a surveillance target behavior, for example, if a zebra crossing in traffic surveillance can be used as a reference live-action for pedestrians and vehicles running red light; in worksite monitoring, construction equipment around workers and a building being constructed can be used as reference reality for dangerous behaviors of the workers (for example, if safety helmets are not worn, safety ropes are not worn, and the like).

In this embodiment, the motion features include a position change feature and an action change feature of the monitored object, where the position change feature refers to a change feature of a relative position between the monitored object and a reference live-action corresponding to the monitored object.

In this embodiment, the abnormal behavior refers to dangerous behavior of the target monitored in the target monitoring area, for example, if the vehicle is parked in the intersection 50 during traffic monitoring; in store monitoring, the act of taking the merchandise directly out of the store without checkout at the checkout counter after the merchandise is taken.

The working principle and the beneficial effects of the embodiment are that: according to the invention, the monitoring videos of multiple angles of the target monitoring area are obtained through multiple shooting devices of the data acquisition module, and the monitoring videos are spliced to generate the panoramic monitoring video, so that the behavior action of each monitoring target in the target monitoring area can be completely seen, and the accuracy of monitoring target behavior identification is effectively improved; the feature extraction module is used for marking the monitoring target on the panoramic monitoring video picture, so that the monitoring equipment can apply various monitoring scenes, synchronous monitoring of various monitoring targets in the same monitoring area is completed, the reference real scene corresponding to the monitoring target is determined, the prejudgment of abnormal behavior related to the monitoring target is completed, then the monitoring target is subjected to action positioning based on the reference real scene, the motion features of the monitoring target are extracted according to the motion positioning, the actual motion track of the monitoring target and the motion behavior relative to the reference real scene can be accurately identified, the accuracy of the operation features of the monitoring target is effectively improved, and the accuracy of the abnormal behavior identification of the judgment early warning module is provided; and judging whether the monitoring target has abnormal behaviors or not through the judging and early warning module based on the motion characteristics, if so, generating a corresponding abnormal early warning notice to send a monitoring management center based on the abnormal behaviors, automatically completing abnormal behavior monitoring in a target monitoring area, reducing the workload of monitoring and checking staff, avoiding errors of manual judgment, and reducing the misjudgment rate of the abnormal behaviors.

Example 2: on the basis of embodiment 1, the data acquisition module, if shown in fig. 2, includes: the installation recommendation sub-module is used for generating a shooting device installation strategy corresponding to each monitoring position in the target monitoring area according to the actual monitoring requirement of the target monitoring area, and comprises the following steps:

the first determining unit is used for obtaining effective monitoring distances corresponding to all monitoring positions of the target monitoring area and determining angle rotation homing time of the shooting equipment based on the scene preset speed and the effective monitoring distances;

a second determining unit, configured to determine a monitoring angle of each photographing apparatus based on the angular rotation homing time and an angular velocity of the camera;

the strategy generation unit is used for determining the target quantity of shooting equipment corresponding to the target monitoring area according to the panoramic monitoring angle corresponding to the monitoring position and the monitoring angle, generating equipment installation recommendation strategies corresponding to the monitoring position and sending the equipment installation recommendation strategies to the monitoring management center;

and the parameter setting unit is used for automatically setting shooting parameters of all shooting equipment according to the monitoring angles of all the shooting equipment and the actual installation positions of the shooting equipment after the shooting equipment is installed.

In this embodiment, the monitoring position refers to the mounting position of the photographing apparatus.

In this embodiment, the effective monitoring distance refers to a distance between a position, which is required to be monitored at the monitoring position and can obtain a clear image, and a nearest position, which can be photographed by the photographing device, where the farthest distance is less than or equal to the effective photographing distance of the photographing device.

In this embodiment, the angular rotation homing time refers to the quotient of the effective monitoring distance and the scene preset speed, wherein the scene preset speed refers to the speed determined according to the scene difference of intelligent video monitoring, for example, the main monitoring target is a person in the monitoring scene of a mall, and the scene preset speed is generally within 2-5 m/s; the method mainly monitors the illegal behaviors of vehicles as targets in traffic monitoring scenes, and the preset speed value of the scenes is generally within 30-60 km/h. Due to different actual monitoring requirements, the preset speeds of the same scene at different positions are different.

In this embodiment, the monitoring angle refers to the product of the angular rotational homing time and the angular velocity of the camera.

In this embodiment, the panoramic monitoring angle refers to a total monitoring angle of a monitoring position of the photographing device in the whole target monitoring area.

In this embodiment, the automatic setting of the shooting parameters refers to setting the rotation speed and the rotation angle of each shooting device in the target monitoring area, so that the rotation periods of the shooting devices in the target monitoring area are the same, wherein the rotation periods refer to the same time from the monitoring angle starting position to the monitoring angle ending position corresponding to the shooting devices, so that the panoramic monitoring video can be formed by splicing even multiple types of shooting devices in the target monitoring area.

The working principle and the beneficial effects of the embodiment are that: according to the invention, the installation recommendation sub-module is used for determining the installation strategy of the shooting equipment corresponding to each monitoring position in the target monitoring area, so that each monitoring position can be fully covered and monitored without dead angles in the corresponding panoramic monitoring angle, and the obtained multi-angle monitoring video without dead angles is ensured.

Example 3: on the basis of embodiment 1, the data acquisition module, if shown in fig. 2, further includes:

the video acquisition unit is used for acquiring monitoring videos of different angles based on a plurality of shooting devices of the target monitoring area;

the sequence determining unit is used for acquiring shooting parameters corresponding to each shooting device in the target monitoring area and determining a splicing sequence of pictures corresponding to the monitoring video based on the shooting parameters;

The video stitching unit is used for stitching pictures of the monitoring video at the same moment according to the stitching sequence to obtain panoramic pictures, and generating panoramic monitoring videos according to the time axis sequence based on the panoramic pictures.

In this embodiment, the splicing sequence refers to a sequence formed by the connection sequence of the monitoring videos at the same time in the monitoring area.

In this embodiment, the panoramic image refers to a panoramic image of a target monitoring area formed by splicing images of monitoring videos of the monitoring videos at the same time.

The working principle and the beneficial effects of the embodiment are that: according to the invention, based on a plurality of shooting devices in the target monitoring area, monitoring videos with different angles are obtained, a splicing sequence of pictures corresponding to the monitoring videos is determined according to shooting parameters corresponding to each shooting device in each target monitoring area, then pictures of the monitoring videos at the same moment are spliced according to the splicing sequence to obtain panoramic pictures, panoramic monitoring videos are generated according to a time axis sequence based on the panoramic pictures, an omnibearing image of the target monitoring area is obtained, the actual situation of each target element in the target monitoring area is completely presented, the extraction accuracy of the running sign of the monitoring target is effectively improved, and the early warning accuracy of an intelligent monitoring system is improved.

Example 4: on the basis of embodiment 3, the video stitching unit includes:

the direction determining subunit is used for respectively determining a monitoring angle starting position and a monitoring angle ending position of each monitoring video, determining a monitoring period corresponding to each monitoring video based on the monitoring angle starting position and the monitoring angle ending position, and generating a forward displacement vector and a reverse displacement vector of the shooting equipment;

the forward splicing unit is used for acquiring a first monitoring sub-video corresponding to the upper half monitoring period, acquiring a first video picture of the first monitoring sub-video frame by frame based on the forward displacement vector, and carrying out picture splicing on the first video picture to acquire a forward picture of the monitoring angle;

acquiring all forward pictures of all shooting devices in a target monitoring area in the same monitoring period, and splicing the forward monitoring pictures according to the splicing sequence to obtain a forward panoramic picture;

the reverse splicing unit is used for acquiring a second monitoring sub-video corresponding to the lower half monitoring period, acquiring a second video picture of the second monitoring sub-video frame by frame based on the reverse displacement vector, and carrying out picture splicing on the second video picture to acquire a reverse picture of the monitoring angle;

Acquiring all reverse images of all shooting devices in a target monitoring area in the same monitoring period, and splicing the reverse monitoring images according to the reverse sequence of the splicing sequence to obtain a reverse panoramic image;

and the panoramic display subunit is used for alternately displaying the forward panoramic picture and the reverse panoramic picture based on the time axis sequence to generate a panoramic monitoring video.

In this embodiment, the monitoring angle starting position refers to an angle corresponding to a shooting starting position of each monitoring video; the monitoring angle end position refers to an angle corresponding to the shooting end position of each monitoring video.

In this embodiment, the monitoring period refers to an angular rotation period of each photographing apparatus, and includes two parts of forward rotation (from a monitoring angle start position to a monitoring angle end position) and reverse rotation (from a monitoring angle end position to a monitoring angle start position). In order to ensure that no dead angle shooting of a target monitoring area can be completed in the same monitoring half period, different forward and reverse directions of rotation of shooting equipment at different monitoring positions are defined differently.

In this embodiment, the upper half of the monitoring period refers to the portion that rotates in the forward direction; the lower half of the monitoring period refers to the portion of the reverse rotation.

In this embodiment, the first monitoring sub-video refers to a monitoring video collected by the shooting device in forward rotation; the second monitoring sub-video is a monitoring video collected by the shooting equipment in a reverse rotation mode.

In this embodiment, the forward displacement vector refers to a vector indicating the forward rotation direction and the total rotation angle of the photographing apparatus; the reverse displacement vector refers to a vector having the same magnitude as the forward displacement vector but opposite direction.

In this embodiment, the first video frame refers to all frames of each frame of the first monitoring sub-video; the second video frame refers to all of each frame of the second monitoring sub-video.

In this embodiment, the forward picture refers to a complete picture of the whole monitoring angle corresponding to the forward rotation of the photographing apparatus; the reverse screen refers to a complete screen of the entire monitoring angle corresponding to the reverse rotation of the photographing apparatus.

In this embodiment, the forward panoramic picture refers to a panoramic picture obtained by stitching monitoring videos collected during forward rotation of all photographing devices in the target monitoring area; the reverse panoramic picture is a panoramic picture obtained by splicing monitoring videos acquired in the process that all shooting devices in the target monitoring area reversely rotate.

In this embodiment, the reverse sequence refers to a new sequence obtained by reversing the splicing sequence.

The working principle and the beneficial effects of the embodiment are that: according to the method, the monitoring angle starting position and the monitoring angle ending position of each monitoring video are respectively determined, the monitoring period corresponding to each monitoring video is determined based on the monitoring angle starting position and the monitoring angle ending position, the forward displacement vector and the reverse displacement vector of the shooting equipment are generated, the direction is determined for splicing the frames of the mixed video of the first monitoring sub video or the second monitoring sub video, the frames are ensured to be connected correctly, the accurate unidirectional panoramic frames are obtained, then the forward panoramic frames and the reverse panoramic frames are alternately displayed based on the time axis sequence, the panoramic monitoring video is generated, and a reliable basis is provided for judging the intelligent abnormal behaviors of the target monitoring area.

Example 5: on the basis of embodiment 1, the feature extraction module, if shown in fig. 3, includes:

the target marking unit is used for determining object characteristics corresponding to the video monitoring object based on the monitoring setting information, carrying out target identification on the panoramic monitoring video based on the object characteristics, and carrying out monitoring target marking according to a target identification result;

The reference determining unit is used for acquiring monitoring safety standards based on the monitoring setting information, determining a safety identification reference object, and carrying out reference marking in the panoramic monitoring video based on the safety identification reference object to obtain a reference live-action;

the reference corresponding unit is used for determining reference live-action corresponding to each monitoring target according to the nearby principle;

the action marking unit is used for coding the monitoring targets based on the monitoring target marks to obtain a monitoring target sequence, and according to the monitoring target sequence, action positioning is carried out on different monitoring targets in the panoramic monitoring video successively, and the motion characteristics of different monitoring targets are extracted by combining the relative relation between the monitoring targets and the corresponding reference live-action.

In this embodiment, the monitoring setting information refers to monitoring information input to the system by a manager according to an actual monitoring requirement of a target monitoring area, and includes information such as a type (e.g., a person, a motor vehicle, a distribution box, etc.) corresponding to a monitoring target and abnormal behavior corresponding to monitoring of each monitoring target.

In this embodiment, the object features refer to object signs corresponding to all monitoring targets corresponding to the target monitoring area.

In this embodiment, the monitoring safety standard refers to a standard of behavior authenticated as safe, for example, the vehicle is not parked on the zebra stripes.

In this embodiment, the security authentication reference refers to a reference that is used as a determination criterion in monitoring security standards, for example, the vehicle is not stopped on a zebra crossing, wherein the zebra crossing can be used as the security authentication reference; the safety helmet must be worn when entering the construction area, and construction equipment on the construction site can be used as a safety identification reference for entering the construction area.

In this embodiment, the monitoring target sequence refers to a sequence including all monitoring target codes on the current panoramic monitoring video picture, so as to avoid missing a certain monitoring target in the feature extraction process.

The working principle and the beneficial effects of the embodiment are that: according to the invention, the object feature corresponding to the video monitoring object is determined based on the monitoring setting information through the object marking unit, the panoramic monitoring video is subjected to object recognition based on the object feature, the monitoring object is marked according to the object recognition result, the monitoring object is automatically recognized according to the actual monitoring requirement, and the recognition of multiple or multiple monitoring objects in the same scene can be completed; and finally, the action marking unit is used for coding the monitoring targets based on the monitoring target marks to obtain monitoring target sequences, and according to the monitoring target sequences, different monitoring targets are successively positioned in the panoramic monitoring video in action, and the motion characteristics of the different monitoring targets are extracted in combination with the relative relation between the monitoring targets and the corresponding reference live-action, so that the actual motion track of the monitoring targets and the motion behavior of the reference live-action can be accurately identified, the accuracy of the operation characteristics of the monitoring targets is effectively improved, and the misjudgment probability of abnormal behavior judgment is reduced.

Example 6: on the basis of embodiment 5, the reference determination unit includes:

the standard acquisition subunit is used for determining a plurality of monitoring contents corresponding to the target monitoring area and standard setting information corresponding to each monitoring content based on the monitoring setting information;

generating corresponding safety standards according to the standard setting information respectively, and acquiring the safety standards corresponding to all monitoring contents to generate monitoring safety standards corresponding to the monitoring positions;

the reference determination subunit is used for determining a plurality of safety identification references corresponding to the monitoring positions according to the monitoring safety standards and acquiring object characteristics corresponding to each safety identification reference;

and the reference calibration unit is used for carrying out reference marking in the panoramic monitoring video based on the object characteristics, and taking the real scene of the reference marking as the reference real scene currently corresponding to the monitoring position.

In this embodiment, the object feature refers to a shape feature or an actual use feature of the security authentication reference.

The working principle and the beneficial effects of the embodiment are that: the invention determines a plurality of monitoring contents corresponding to the target monitoring area and standard setting information corresponding to each monitoring content through the standard acquisition subunit based on the monitoring setting information, thereby realizing the setting of a plurality of monitoring targets of the intelligent monitoring system, and then can complete the judgment and early warning of the plurality of monitoring targets on the same picture, effectively provide the monitoring utilization rate, generate corresponding safety standards according to the standard setting information, acquire the safety standards corresponding to all the monitoring contents and generate the monitoring safety standards corresponding to the monitoring positions, complete the monitoring standards of various monitoring targets in the target monitoring area, and provide basis for the selection of reference scenes and the judgment of abnormal behaviors; and the reference determination subunit determines a plurality of safety identification references corresponding to the monitoring positions according to the monitoring safety standards, acquires object characteristics corresponding to each safety identification reference, rapidly performs reference marks in the panoramic monitoring video based on the object characteristics through the reference calibration unit, and takes the real scenes of the reference marks as the reference real scenes currently corresponding to the monitoring positions, thereby being beneficial to rapidly determining the matching corresponding relation between the monitoring targets and each reference real scene.

Example 7: on the basis of embodiment 5, the action marking unit includes:

the marking subunit is used for marking different monitoring targets in the video frames corresponding to the current panoramic monitoring video according to the target monitoring sequence to obtain a first action, and comparing the first action position with the reference live-action position to obtain a first relative relation;

the acquisition subunit is used for acquiring second actions corresponding to different monitoring targets on a previous frame picture of a video frame corresponding to the current panoramic monitoring video and second relative relations obtained by comparing the second action positions with reference live-action positions according to the target monitoring sequence;

and the comparison subunit is used for comparing the first action with the second action to obtain action change characteristics, comparing the first relative relation with the second relative relation to obtain position change characteristics, and generating the movement characteristics of the monitoring target based on the action change characteristics and the position change characteristics.

In this embodiment, the first action refers to determining an action gesture by performing an action mark on a monitoring target in a video frame corresponding to the current panoramic monitoring video according to a preset action mark key point; and the second action is to perform action marking on the monitoring target in the reverse video frame corresponding to the current panoramic monitoring video according to the preset action marking key point to determine the action gesture.

In this embodiment, the first relative relationship refers to a relative angle and a distance relationship between the first action position and the reference live-action position; the second relative relationship refers to a relative angle and distance relationship between the second action position and the reference live-action position.

In this embodiment, the reverse video frame refers to a video frame with a rotation direction opposite to that of the shooting device of the video frame corresponding to the current panoramic surveillance video, that is, a frame previous to the video frame corresponding to the current panoramic surveillance video.

In this embodiment, the motion change feature refers to a motion change condition of the same monitoring target in adjacent video frames.

In this embodiment, the location change feature refers to a location change condition of the same monitoring target and the same reference live-action corresponding to the same monitoring target in adjacent video frames.

The working principle and the beneficial effects of the embodiment are that: the method comprises the steps that different monitored first actions and first relative relations in video frames corresponding to a current panoramic monitoring video are obtained through a marking subunit; and then, obtaining a second action and a second relative relation of adjacent video frames corresponding to the current panoramic monitoring video through the obtaining subunit, comparing the first action with the second action, the first relative relation with the second relative relation through the comparing subunit, respectively obtaining action change characteristics and position change characteristics to obtain the motion characteristics of the monitoring target, and fusing the motion trend change characteristics into the current motion behavior characteristics of the monitoring target, so that timeliness of the motion characteristics is enhanced, the system can update reference scenes according to actual running conditions of the monitoring target conveniently, and the accuracy of abnormal behavior judgment is improved.

Example 8: on the basis of embodiment 7, the reference determination unit further includes:

the reference updating subunit is used for updating the reference live-action corresponding to the monitoring target based on the nearby principle after the first relative relation of the video frame corresponding to the current panoramic monitoring video, determining the current live-action of the monitoring target, taking the first action as the second action when the current reference live-action is consistent with the original reference live-action, taking the first relative relation as the second relative relation, and waiting for the call of the acquisition subunit;

otherwise, comparing the current reference live-action position with the first action position, obtaining a second relative relation according to a comparison result, and taking the first action as a second action to wait for the call of the acquisition subunit.

The working principle and the beneficial effects of the embodiment are that: after a first relative relation of a video frame corresponding to a current panoramic monitoring video, updating a reference live-action of a monitoring target based on a nearby principle, determining the current live-action of the monitoring target, taking the first action as a second action when the current reference live-action is consistent with an original reference live-action, taking the first relative relation as a second relative relation, and waiting for the call of a subunit to be acquired; otherwise, comparing the current reference live-action position with the first action position, obtaining a second relative relation according to a comparison result, taking the first action as the second action, waiting for the call of the acquisition subunit, ensuring that the reference live-action of the comparison action by the action marking unit is consistent, ensuring the accuracy of judging the abnormal behavior of the video frame picture corresponding to the current panoramic monitoring video, and updating the corresponding reference live-action according to the actual motion of the monitoring target in time, thereby achieving the aim of updating the corresponding predicted abnormal behavior along with the motion of the monitoring target, and realizing the judgment of multiple behaviors of the monitoring target.

Example 9: on the basis of embodiment 1, the judgment early warning module, if shown in fig. 4, includes:

the standard determining unit is used for determining all the corresponding to-be-selected abnormal behaviors of different monitoring targets in the target monitoring area and the abnormal motion characteristics corresponding to the to-be-selected abnormal behaviors according to the monitoring setting information, and determining the corresponding target abnormal behaviors of the different monitoring targets and the corresponding target abnormal motion characteristics based on the reference real scenes corresponding to the different monitoring targets;

the characteristic comparison unit is used for comparing the motion characteristics of different monitoring targets with the corresponding abnormal motion characteristics of the targets to obtain the characteristic similarity;

the judging and early warning unit is used for judging that the monitoring target behavior is abnormal when the feature similarity is larger than or equal to a preset value, and generating a corresponding abnormal early warning notice according to the target abnormal behavior corresponding to the monitoring target to send a monitoring management center;

and when the feature similarity is smaller than a preset value, judging that the monitoring target acts normally, and continuously monitoring the monitoring target.

In this embodiment, the abnormal behavior to be selected refers to all behaviors of the monitoring target considered to be abnormal.

In this embodiment, the abnormal motion feature refers to an operation feature corresponding to each abnormal behavior to be selected of the monitoring target by using a reference object of the corresponding reference live-action.

In this embodiment, the feature similarity refers to the similarity between the current motion feature of the monitored target and the corresponding abnormal motion feature of the target.

The working principle and the beneficial effects of the embodiment are that: according to the invention, all the to-be-selected abnormal behaviors corresponding to different monitoring targets in the target monitoring area and the abnormal motion characteristics corresponding to the to-be-selected abnormal behaviors are determined through the standard determining unit according to the monitoring setting information, the target abnormal behaviors corresponding to the different monitoring targets and the target abnormal motion characteristics corresponding to the different monitoring targets are determined based on the reference live-action corresponding to the different monitoring targets, the prediction of the abnormal behaviors is completed according to the reference live-action, the abnormal behavior judging step is simplified, and the judging result output speed of the abnormal behaviors is effectively accelerated; the feature comparison unit is used for comparing the motion features of different monitoring targets with the corresponding abnormal motion features of the targets to obtain feature similarity, so that the problem of misjudgment caused by the difference of motion feature acquisition caused by individual errors is avoided, and the accuracy of abnormal behavior judgment is improved; when the feature similarity is larger than or equal to a preset value, the judgment early warning unit judges that the monitoring target behavior is abnormal, and generates a corresponding abnormal early warning notification according to the target abnormal behavior corresponding to the monitoring target to send a monitoring management center; when the feature similarity is smaller than a preset value, the monitoring target is judged to be normal in behavior, and the monitoring target is continuously monitored, so that intelligent early warning of an intelligent monitoring video is completed, the working pressure of monitoring and checking staff is reduced, the problem of manual misjudgment is avoided, and the abnormal behavior of a target monitoring area is ensured to be found in time.

Example 10: on the basis of embodiment 1, an intelligent video monitoring system further includes: the remote control module is used for remotely controlling video monitoring by the monitoring management center to finish verifying abnormal behavior early warning, and comprises the following steps:

the target selecting unit is used for selecting an amplified viewing target in the panoramic monitoring video by the monitoring management center;

the target determining unit is used for determining a panoramic display subarea where the amplified reference target is located, determining a monitoring position and a corresponding monitoring angle based on the panoramic display subarea, and determining target shooting equipment according to the monitoring position and the corresponding monitoring angle;

the focusing and amplifying unit is used for controlling and adjusting the shooting focal length of the target shooting equipment to focus and shoot the amplified viewing target so as to obtain a local amplified video;

and the sending display unit is used for sending the local amplified video to a preset display position for local viewing for display.

In this embodiment, the panoramic display sub-area refers to a sub-area where a zoom-in reference target is displayed in the panoramic video, where different monitoring positions correspond to one display sub-area. There are multiple monitoring locations for the target monitoring area.

In the present embodiment, the target photographing apparatus refers to a photographing apparatus that photographs an enlarged view target.

In this embodiment, the locally enlarged video refers to a monitoring video obtained by focusing shooting of an enlarged view target.

In this embodiment, the preset display position refers to a display area reserved by the monitoring management center for the locally enlarged video.

The working principle and the beneficial effects of the embodiment are that: according to the invention, the remote control module monitors the current actual condition of a certain position in the target monitoring area through the remote control and accurate check of the local monitoring area or the amplification and accurate check of a certain monitoring target in the target monitoring area, so that the remote abnormal behavior verification by the monitoring and management center is facilitated. The monitoring management center selects an amplified viewing target in the panoramic monitoring video through the target selection unit, then determines a panoramic display subarea where the amplified viewing target is located through the target determination unit, determines a monitoring position and a corresponding monitoring angle based on the panoramic display subarea, determines target shooting equipment according to the monitoring position and the corresponding monitoring angle, and controls and adjusts the shooting focal length of the target shooting equipment to focus the amplified viewing target through the focusing amplification unit to obtain a local amplified video, so that accurate shooting of the amplified viewing target is realized, interference of other irrelevant targets on local amplification is reduced, and the accuracy of remote abnormal behavior verification of the monitoring management center is improved; the local amplified video is sent to the preset display position for local viewing through the sending display unit to be displayed, so that the display of the panoramic monitoring video is not influenced while the local amplified viewing is ensured, the real-time display of the panoramic monitoring video is ensured, and the monitoring management center can timely know the real-time dynamic information of the target monitoring area.

It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.

Claims (9)

1. An intelligent video monitoring system, comprising:

the data acquisition module is used for acquiring monitoring videos of multiple angles of a target monitoring area based on multiple shooting devices, and splicing the monitoring videos to generate a panoramic monitoring video;

the feature extraction module is used for marking a monitoring target on the panoramic monitoring video, determining a reference live-action corresponding to the monitoring target, performing action positioning on the monitoring target based on the reference live-action, and extracting the motion features of the monitoring target according to the motion positioning;

the judging and early warning module is used for judging whether the monitoring target has abnormal behaviors or not based on the motion characteristics, and if so, generating corresponding abnormal early warning notification and sending the monitoring management center based on the abnormal behaviors of the monitoring target;

the installation recommendation sub-module is configured to generate a shooting device installation policy corresponding to each monitoring position in the target monitoring area according to an actual monitoring requirement of the target monitoring area, and includes:

The first determining unit is used for obtaining effective monitoring distances corresponding to all monitoring positions of the target monitoring area and determining angle rotation homing time of the shooting equipment based on the scene preset speed and the effective monitoring distances;

a second determining unit, configured to determine a monitoring angle of each photographing apparatus based on the angular rotation homing time and an angular velocity of the camera;

the strategy generation unit is used for determining the target quantity of shooting equipment corresponding to the target monitoring area according to the panoramic monitoring angle corresponding to the monitoring position and the monitoring angle, generating equipment installation recommendation strategies corresponding to the monitoring position and sending the equipment installation recommendation strategies to the monitoring management center;

and the parameter setting unit is used for automatically setting shooting parameters of all shooting equipment according to the monitoring angles of all the shooting equipment and the actual installation positions of the shooting equipment after the shooting equipment is installed.

2. The intelligent video monitoring system of claim 1, wherein the data acquisition module further comprises:

the video acquisition unit is used for acquiring monitoring videos of different angles based on a plurality of shooting devices of the target monitoring area;

the sequence determining unit is used for acquiring shooting parameters corresponding to each shooting device in the target monitoring area and determining a splicing sequence of pictures corresponding to the monitoring video based on the shooting parameters;

The video stitching unit is used for stitching pictures of the monitoring video at the same moment according to the stitching sequence to obtain panoramic pictures, and generating panoramic monitoring videos according to the time axis sequence based on the panoramic pictures.

3. The intelligent video monitoring system of claim 2, wherein the video stitching unit comprises:

the direction determining subunit is used for respectively determining a monitoring angle starting position and a monitoring angle ending position of each monitoring video, determining a monitoring period corresponding to each monitoring video based on the monitoring angle starting position and the monitoring angle ending position, and generating a forward displacement vector and a reverse displacement vector of the shooting equipment;

the forward splicing unit is used for acquiring a first monitoring sub-video corresponding to the upper half monitoring period, acquiring a first video picture of the first monitoring sub-video frame by frame, and carrying out picture splicing on the first video picture based on a forward displacement vector to acquire a forward picture of the monitoring angle;

acquiring all forward pictures of all shooting devices in a target monitoring area in the same monitoring period, and splicing the forward monitoring pictures according to the splicing sequence to obtain a forward panoramic picture;

The reverse splicing unit is used for acquiring a second monitoring sub-video corresponding to the lower half monitoring period, acquiring a second video picture of the second monitoring sub-video frame by frame, and carrying out picture splicing on the second video picture based on a reverse displacement vector to acquire a reverse picture of the monitoring angle;

acquiring all reverse images of all shooting devices in a target monitoring area in the same monitoring period, and splicing the reverse monitoring images according to the reverse sequence of the splicing sequence to obtain a reverse panoramic image;

and the panoramic display subunit is used for alternately displaying the forward panoramic picture and the reverse panoramic picture based on the time axis sequence to generate a panoramic monitoring video.

4. The intelligent video monitoring system of claim 1, wherein the feature extraction module comprises:

the target marking unit is used for determining object characteristics corresponding to the video monitoring object based on the monitoring setting information, carrying out target identification on the panoramic monitoring video based on the object characteristics, and carrying out monitoring target marking according to a target identification result;

the reference determining unit is used for acquiring monitoring safety standards based on the monitoring setting information, determining a safety identification reference object, and carrying out reference marking in the panoramic monitoring video based on the safety identification reference object to obtain a reference live-action;

The reference corresponding unit is used for determining reference live-action corresponding to each monitoring target according to the nearby principle;

the action marking unit is used for coding the monitoring targets based on the monitoring target marks to obtain a monitoring target sequence, and according to the monitoring target sequence, action positioning is carried out on different monitoring targets in the panoramic monitoring video successively, and the motion characteristics of different monitoring targets are extracted by combining the relative relation between the monitoring targets and the corresponding reference live-action.

5. The intelligent video monitoring system of claim 4, wherein the reference determination unit comprises:

the standard acquisition subunit is used for determining a plurality of monitoring contents corresponding to the target monitoring area and standard setting information corresponding to each monitoring content based on the monitoring setting information;

generating corresponding safety standards according to the standard setting information respectively, and acquiring the safety standards corresponding to all monitoring contents to generate monitoring safety standards corresponding to the monitoring positions;

the reference determination subunit is used for determining a plurality of safety identification references corresponding to the monitoring positions according to the monitoring safety standards and acquiring object characteristics corresponding to each safety identification reference;

And the reference calibration unit is used for carrying out reference marking in the panoramic monitoring video based on the object characteristics, and taking the real scene of the reference marking as the reference real scene currently corresponding to the monitoring position.

6. The intelligent video monitoring system of claim 4, wherein the action marking unit comprises:

the marking subunit is used for marking different monitoring targets in the video frames corresponding to the current panoramic monitoring video according to the target monitoring sequence to obtain a first action, and comparing the first action position with the reference live-action position to obtain a first relative relation;

the acquisition subunit is used for acquiring second actions and second relative relations corresponding to different monitoring targets on a previous frame picture of a video frame corresponding to the current panoramic monitoring video according to the target monitoring sequence;

and the comparison subunit is used for comparing the first action with the second action to obtain action change characteristics, comparing the first relative relation with the second relative relation to obtain position change characteristics, and generating the movement characteristics of the monitoring target based on the action change characteristics and the position change characteristics.

7. The intelligent video monitoring system of claim 6, wherein the reference determination unit further comprises:

The reference updating subunit is used for updating the reference live-action corresponding to the monitoring target based on the nearby principle after the first relative relation of the video frame corresponding to the current panoramic monitoring video, determining the current live-action of the monitoring target, taking the first action as the second action when the current reference live-action is consistent with the original reference live-action, taking the first relative relation as the second relative relation, and waiting for the call of the acquisition subunit;

otherwise, comparing the current reference live-action position with the first action position, obtaining a second relative relation according to a comparison result, and taking the first action as a second action to wait for the call of the acquisition subunit.

8. The intelligent video monitoring system according to claim 1, wherein the judgment pre-warning module comprises:

the standard determining unit is used for determining all the corresponding to-be-selected abnormal behaviors of different monitoring targets in the target monitoring area and the abnormal motion characteristics corresponding to the to-be-selected abnormal behaviors according to the monitoring setting information, and determining the corresponding target abnormal behaviors of the different monitoring targets and the corresponding target abnormal motion characteristics based on the reference real scenes corresponding to the different monitoring targets;

The characteristic comparison unit is used for comparing the motion characteristics of different monitoring targets with the corresponding abnormal motion characteristics of the targets to obtain the characteristic similarity;

the judging and early warning unit is used for judging that the monitoring target behavior is abnormal when the feature similarity is larger than or equal to a preset value, and generating a corresponding abnormal early warning notice according to the target abnormal behavior corresponding to the monitoring target to send a monitoring management center;

and when the feature similarity is smaller than a preset value, judging that the monitoring target acts normally, and continuously monitoring the monitoring target.

9. The intelligent video monitoring system of claim 1, further comprising: the remote control module is used for monitoring the remote control of the management center and accurately checking the current actual condition of a certain position in the target monitoring area, and comprises the following steps:

the target selecting unit is used for selecting an amplified viewing target in the panoramic monitoring video by the monitoring management center;

the target determining unit is used for determining a panoramic display subarea where the amplified reference target is located, determining a monitoring position and a corresponding monitoring angle based on the panoramic display subarea, and determining target shooting equipment according to the monitoring position and the corresponding monitoring angle;

The focusing and amplifying unit is used for controlling and adjusting the shooting focal length of the target shooting equipment to focus and shoot the amplified viewing target so as to obtain a local amplified video;

and the sending display unit is used for sending the local amplified video to a preset display position for local viewing for display.