JP7571960B1 - PROGRAM, INFORMATION PROCESSING APPARATUS, INFORMATION PROCESSING METHOD, AND LEARNING MODEL - Google Patents

Abstract

[Problem] To generate data obtained by observing an observation area using multiple flying objects at a high frequency. [Solution] A program that causes a computer to execute the following: based on first observation data, which is information on the observation area at a first timing and observed by a first observation means, and second observation data, which is information on the observation area at a second timing different from the first timing and observed by a second observation means different from the first observation means, generate integrated data including first information based on the first observation data and second information based on the second observation data. [Selected Figure] Figure 2

Description

The present invention relates to a program, an information processing device, an information processing method, and a learning model.

Observations of the Earth's surface, including land and sea, are widely carried out using flying objects such as artificial satellites, aircraft, and drones. Observation methods using flying objects include those that acquire optical images, and those that acquire radar images obtained using synthetic aperture radar (SAR) technology, known as SAR images.

Patent Document 1 describes a method for grasping the conditions on the Earth's surface that detects changed areas by aligning and comparing a new image obtained by photographing a specific observation point on the Earth's surface with a first flying object flying on a first orbit with a most recent previous image obtained by photographing the same point as the specific observation point with a second flying object flying on a second orbit different from the first orbit, in order to improve the frequency with which the conditions on the Earth's surface are grasped.

JP 2022-167343 A

In the method described in Patent Document 1, new images observed from a first orbit are treated as separate images from past images observed from a second orbit. In Patent Document 1, new images are new images observed by a first flying object from a first orbit, and are not data that takes into account observation of the observation point by a second flying object. Therefore, in Patent Document 1, the data itself of the observation point is not generated frequently, and only the frequency of change detection is improved. When observations are made using flying objects, it is preferable to generate observation data frequently in order to grasp the state of the observed area in more detail.

The present invention aims to generate data obtained by observing an observation area using multiple flying objects at a high frequency.

A program according to one aspect of the present invention causes a computer to execute the following: based on first observation data, which is information on an observation area at a first timing and observed by a first observation means, and second observation data, which is information on an observation area at a second timing different from the first timing and observed by a second observation means different from the first observation means, generate integrated data including first information based on the first observation data and second information based on the second observation data.

According to this aspect, the integrated data includes first information based on the first observation data and second information based on the second observation data, so that when at least one of the first observation data or the second observation data is updated or acquired, the integrated data is also updated. Thus, the computer can generate integrated data frequently by generating integrated data based on information of the observation area at either the first timing or the second timing.

In the above aspect, the first observation data is data observed under the first observation conditions, and the computer acquires third observation data, which is information on the observation area at a third timing different from the first timing, which is data observed under the third observation conditions by a third observation means different from the first observation means, and which is observed based on a common observation method with the first observation data; and inputs the third observation data, the third observation conditions, and the third timing into a conversion model that receives input of the learning observation data, which is information on the learning observation area, the learning observation conditions corresponding to the learning observation data, and the learning observation timing at which the learning observation data was observed, and outputs converted observation data in which the learning observation data is converted, and which indicates data observed in the learning observation area under the first observation conditions at the learning observation timing, and further executes the above steps, and generating the integrated data may include updating the first information based on the converted observation data to generate the integrated data.

According to this aspect, when there is third observation data that has a common observation method with the first observation data, the third observation data can be converted into data observed under the first observation conditions. Therefore, for example, when performing an analysis using the integrated data, it becomes possible to treat data observed under different observation conditions as data observed under the first observation conditions, that is, as the first information, and the first information in the integrated data is updated frequently, and as a result, the integrated data can also be generated frequently.

In the above aspect, the first observation condition may include first aircraft information of a first flying object, which is the first observation means, and first environmental information indicating the observation environment by the first flying object, and the third observation condition may include third aircraft information of a third flying object, which is the third observation means, and third environmental information indicating the observation environment by the third flying object.

According to this aspect, data conversion is possible taking into account the aircraft information and the observation environment. Therefore, it is possible to convert the third observation data into observation data under the first observation conditions with high accuracy.

In the above aspect, the first observation data is data observed under first observation conditions, and the first observation data and the first observation conditions are acquired by a computer, and third observation data is information on the observation domain at a third timing different from the first timing, and is data observed under third observation conditions by a third observation means different from the first observation means, and is observed based on a common observation method with the first observation data. The learning observation data, which is information on the learning observation domain, the learning observation conditions corresponding to the learning observation data, and the learning observation timing at which the learning observation data was observed are input, and converted observation data is obtained by converting the learning observation data, and the learning observation timing is acquired. The method further includes inputting the first observation data, the first observation condition, and the first timing into a conversion model that outputs converted observation data indicating data observed in an observation region under a reference observation condition, and acquiring the first converted observation data as the converted observation data into which the first observation data is converted; inputting the third observation data, the third observation condition, and the third timing into the conversion model, and acquiring the second converted observation data as the converted observation data into which the third observation data at the third timing is converted; and generating the integrated data may include updating the first information based on the first converted observation data with the first information based on the second converted observation data to generate the integrated data.

According to this aspect, each of the first observation data and the third observation data observed under a common observation method can be converted into converted observation data observed under reference observation conditions that are common observation conditions. By generating integrated data based on the converted observation data, the differences between the data in the integrated data can be reduced, and the burden of information processing when analyzing the integrated data can be reduced. In addition, the first information in the integrated data is updated frequently, and as a result, the integrated data can also be generated frequently.

In the above aspect, the first observation condition may include first aircraft information of the first flying object, which is the first observation means, and first environmental information indicating the observation environment by the first flying object, and the third observation condition may include second aircraft information of the third flying object, which is the third observation means, and second environmental information indicating the observation environment by the third flying object.

According to this aspect, data conversion is possible taking into account the aircraft information and the observation environment. Therefore, it is possible to convert the second observation data into observation data under the first observation conditions with high accuracy.

In the above aspect, the first observation data may be data observed based on a first observation method, and the second observation data may be data observed based on a second observation method that is different from the first observation method.

According to this aspect, data observed based on different observation methods are generated as integrated data. For example, the first observation method can be an observation method using SAR, and the second observation method can be an observation method using optical images. This makes it possible to integrate data observed using multiple observation methods into one piece of data, enabling the acquisition of more information through analysis or high-precision analysis.

In the above aspect, the computer may further execute the steps of acquiring third observation data, which is information on the observation area at a third timing different from the first timing, which is data observed under third observation conditions by a third observation means different from the first observation means, and which is observed based on an observation method different from the first observation method, and inputting the third observation data and the third timing into an inference model that receives the learning observation data and the learning observation timing at which the learning observation data was observed, and outputs inferred observation data indicating data in which the observation area was observed based on the first observation method at the learning observation timing, and acquiring inferred observation data at the third timing, and generating the integrated data may include updating the first information based on the inferred observation data to generate the integrated data.

According to this aspect, the observation data at the third timing based on the first observation method can be inferred from the third observation data at the third timing based on the second observation method. As a result, even if the first observation data cannot be obtained due to factors such as bad weather, the data can be supplemented with the third observation data, and integrated data including data observed based on the first observation method can be generated frequently.

In the above aspect, the computer may further execute the steps of acquiring first observation data, inputting the first observation data, the first timing, and the second timing to an inference model that receives input of the learning observation data, a learning observation timing at which the learning observation data was observed, and a timing different from the learning observation timing, and outputs inferred observation data indicating data at which the observation area was observed based on a second observation method at the different timing, and acquiring inferred observation data at the second timing, and generating integrated data may include generating integrated data including second information based on the inferred observation data at the second timing.

According to this aspect, the observation data at the second timing based on the second observation method can be inferred from the first observation data at the first timing based on the first observation method. As a result, even if the second observation data cannot be obtained due to factors such as bad weather, for example, it is possible to supplement the data with the first observation data to generate integrated data.

In the above aspect, the computer may further execute the steps of acquiring third observation data when the observation area is observed at a third timing prior to the first timing, inputting the learning observation data, the learning observation timing at which the learning observation data was observed, and a prediction timing after the learning observation timing into an estimation model that receives inputs of the third observation data, the third timing, and the first timing and acquiring inferred observation data at the first timing, and generating integrated data may include generating integrated data including the first information based on the inferred observation data at the first timing.

According to this aspect, the observation data at the first timing based on the first observation method can be inferred from the third observation data at the third timing based on the first observation method. As a result, even if the first observation data cannot be obtained due to factors such as bad weather, for example, the data can be supplemented with the third observation data observed at the third timing prior to the first timing to generate integrated data.

In the above aspect, the computer may further be caused to receive learning integrated data based on learning observation data, which is information on the learning observation domain, and input the integrated data to a meta-information output model that outputs meta-information on the learning observation domain to obtain meta-information on the observation domain.

According to this aspect, meta-information of the observed area can be obtained using integrated data generated at a high frequency based on observations of the observed area. This makes it possible to improve the accuracy of the meta-information of the observed area.

An information processing device according to another aspect of the present invention includes an integrated data generation unit that generates integrated data including first information based on the first observation data and second information based on the second observation data, based on first observation data, which is information of an observation area at a first timing and observed by a first observation means, and second observation data, which is information of an observation area at a second timing different from the first timing and observed by a second observation means different from the first observation means.

An information processing method according to another aspect of the present invention includes a computer generating integrated data including first information based on the first observation data and second information based on the second observation data, based on first observation data, which is information of an observation area at a first timing and observed by a first observation means, and second observation data, which is information of an observation area at a second timing different from the first timing and observed by a second observation means different from the first observation means.

A learning model according to another aspect of the present invention is trained using teacher data that receives as input learning observation data, which is information about a learning observation area, and learning observation conditions corresponding to the learning observation data, and outputs first converted observation data in which the learning observation data is converted as data observed in the learning observation area according to standard observation conditions at the time when the learning observation data is acquired, and causes a computer to function in response to input of observation data, which is information about a target observation area, and observation conditions corresponding to the observation data, outputting second converted observation data in which the observation data is converted, indicating data observed in the target observation area under standard observation conditions at the time when the observation data is acquired.

A learning model according to another aspect of the present invention is trained using teacher data that receives as input learning observation data when a learning observation area is observed based on a first observation method, an observation timing at which the learning observation data was observed, and a predicted timing different from the observation timing, and outputs first estimated observation data that indicates data observed when the learning observation area is observed based on a second observation method different from the first observation method at the predicted timing, and causes a computer to function in response to input of observation data that is information about the target observation area and observation conditions corresponding to the observation data, outputting second estimated observation data that indicates data observed when the target observation area is observed based on the second observation method at the timing when the observation data was acquired.

The present invention makes it possible to generate data obtained by observing an observation area using multiple flying objects at a high frequency.

FIG. 1 is a schematic diagram of an observation system 10. FIG. 2 is a diagram illustrating generation of integrated data according to the first embodiment. FIG. 1 is a block diagram of an observation data processing device 101 according to a first embodiment. FIG. 2 is a diagram illustrating a meta information output model according to the first embodiment. 10 is a flowchart of an integrated data generation process according to the first embodiment. 11 is a flowchart of a meta information acquisition process according to the first embodiment. FIG. 11 is a block diagram of an observation data processing device 101A according to a second embodiment. FIG. 11 is a diagram illustrating a first conversion model according to the second embodiment. 13 is an example of a flowchart of an integrated data generation process according to the second embodiment. FIG. 11 is a diagram illustrating an example of generation of integrated data according to the second embodiment. FIG. 11 is a diagram illustrating a second conversion model according to the second embodiment. 13 is another example of a flowchart of the integrated data generation process according to the second embodiment. FIG. 13 is a diagram illustrating another example of generation of integrated data according to the second embodiment. FIG. 11 is a block diagram of an observation data processing device 101B according to a third embodiment. FIG. 13 is a diagram illustrating a first estimation model according to the third embodiment. 13 is an example of a flowchart of an integrated data generation process according to the third embodiment. FIG. 13 is a diagram illustrating generation of integrated data according to the third embodiment. FIG. 13 is a diagram illustrating a second estimation model according to the third embodiment. 13 is another example of a flowchart of the integrated data generation process according to the third embodiment. FIG. 13 is a diagram illustrating a third estimation model according to the third embodiment. 13 is another example of a flowchart of the integrated data generation process according to the third embodiment. FIG. 13 is a diagram illustrating generation of integrated data according to the third embodiment.

A preferred embodiment of the present invention will be described with reference to the attached drawings. In each drawing, the same reference numerals are used to denote the same or similar configurations.

First Embodiment
1 shows an observation system 10 according to a first embodiment. The observation system 10 is similar to that of a second embodiment and a third embodiment.

The observation system 10 has an observation data processing device 101, satellites 201, 202, and 203, and an observation data receiving device R.

In the observation system 10, observation data obtained by observing the Earth's surface O using multiple satellites including artificial satellites 201, 202, and 203 is transmitted to the observation data receiving device R. At this time, observation conditions indicating the environment when the observation data is observed may be associated with or included in the observation data and transmitted to the observation data receiving device R. The observation data processing device 101 acquires the observation data from the observation data receiving device R and performs information processing on the observation data. Note that the device that observes the observation data in the observation system 10 is an airborne object that is placed in space as an artificial satellite capable of acquiring observation data and orbits the Earth, as exemplified by the artificial satellites 201, 202, and 203. The airborne object that observes the observation data may be a geostationary satellite. The airborne object may be an aircraft, helicopter, drone, or other device that can be located above the Earth.

In the observation system 10, the observation data observed by the satellites 201, 202, and 203 may be data observed based on different observation methods, or may be data observed based on a common observation method.

The observation method is, for example, a method using SAR signals (SAR method), and the observation data is a signal corresponding to the electromagnetic waves (electromagnetic waves) irradiated to the observation target from an artificial satellite equipped with a radar device and reflected by the observation target. The observation data based on the SAR method (SAR data) is, for example, visualized by the observation data processing device 101 to become a SAR image. As another type of SAR data, there is second level SAR data in which the observed SAR data is subjected to range compression and multi-look azimuth compression. The second level SAR data makes it possible to visualize the SAR image in a geometrically corrected state. As another type of SAR data, there is third level SAR data in which the observed SAR data is subjected to range compression, single-look azimuth compression, and orthorectification. By adding orthorectification, it is possible to obtain a SAR image that can be overlaid on the optical image described below.

Another observation method is the optical method, which uses optical data obtained using optical sensors. Optical data is information obtained by sensors that observe light with wavelengths in the visible light range or near-infrared range, for example. Optical data is data based on light generated when sunlight is reflected by the earth's surface.

Another observation method is to use meteorological data obtained using an infrared sensor. Meteorological data is a signal obtained by observing infrared radiation emitted from clouds, the earth's surface, and the atmosphere, and is used to generate infrared images and water vapor images. In the first and subsequent embodiments, optical data and meteorological data are described separately based on the differences in the wavelength ranges used.

The generation of integrated data by the observation data processing device 101 will be described with reference to FIG. 2. In FIG. 2, an example will be described in which each of the artificial satellites 201, 202, and 203 performs observation based on a mutually different observation method, method 1, method 2, or method 3. Here, for example, method 1 is a method that uses SAR data, method 2 is a method that uses optical data, and method 3 is a method that uses meteorological data.

At a certain time T1 on the time axis T, the artificial satellite 201 observes the observation area and acquires observation data D11. The observation data D11 is transmitted to the observation data processing device 101, which generates integrated data ID1 based on the observation data D11.

Then, the artificial satellite 202 observes observation data D22 at a certain time T2 on the time axis T. The observation data D22 is transmitted to the observation data processing device 101, which generates integrated data ID2 based on the previously observed observation data D11 and the observation data D22.

Integrated data ID2 is data based on observation data based on method 1 at time T1 and observation data based on method 2 at time T2, and is data that contains the most recent information on the observation area at time T2.

The artificial satellite 203 observes observation data D33 at time T3 on the time axis T. The observation data D33 is transmitted to the observation data processing device 101, which generates integrated data ID3 based on the previously observed observation data D11, D22, and observation data D33. The integrated data ID3 is data based on the observation data based on method 1 at time T1, the observation data based on method 2 at time T2, and the observation data based on method 3 at time T3, and is data that includes the most recent information on the observation area at time T3.

Then, at time T4, the artificial satellite 201 observes the same observation area as the area observed at time T1, and obtains observation data D14. Here, the time between observations of the observation area by the artificial satellite 201 is an interval determined, for example, according to the orbit of the artificial satellite 201. The observation data D14 is transmitted to the observation data processing device 101, which generates integrated data ID4 based on the observation data D14. The observation data processing device 101 replaces the observation data D11 obtained by the observation by the artificial satellite 201 at time T1 with the newly obtained observation data D14 to generate integrated data ID14. The integrated data ID4 is data containing the most recent information on the observation area at time T4.

Similarly, at time T5, the artificial satellite 202 observes the same observation area as the area observed at time T2 and obtains observation data D25. The observation data D25 is transmitted to the observation data processing device 101, which generates integrated data ID5 based on the observation data D25. The observation data processing device 101 replaces the observation data D22 obtained by observation by the artificial satellite 202 at time T2 with the newly obtained observation data D25 to generate integrated data ID5. The integrated data ID5 is data that includes the most recent information about the observation area at time T5. By integrating the data at each time, the observation data processing device 101 can generate data obtained by observing an observation area using multiple flying objects more frequently than observation using a single flying object.

The information included in the integrated data is associated with information on the time when the observation data on which each piece of information is based was observed, and the information in the integrated data can be rearranged based on time. For example, in integrated data ID5, it is possible to order the observation data D33 observed at time T3, the observation data D14 observed at time T4, and the observation data D25 observed at time T5 in chronological order. Note that the observation data may not include time information and may be rearranged in the order of acquisition. In other words, the integrated data is generated in a state in which it is possible to rearrange the data in chronological order. By using the integrated data generated in such a way that it is possible to rearrange the observation data in chronological order, it is possible to more accurately detect changes in the observation area. Note that the integrated data does not necessarily have to be generated in a manner in which it is possible to rearrange the observation data in chronological order. The format of the integrated data can be set appropriately depending on the use of the integrated data.

Here, the data to be integrated may be the signals themselves from each satellite, or image data generated based on each signal. When integrated data is generated based on image data, each image data is integrated as a respective channel input to a meta-information output model described below. The data to be integrated may also be meta-information obtained from the signals or image data from each satellite. The signals, image data, or meta-information from each satellite may also be information generated by, for example, a simulation process performed by an information processing device based on observation data. In other words, the data included in the integrated data may be information based on observation data.

The observation data processing device 101 generates integrated data each time it acquires observation data from each satellite. As a result, the integrated data is updated to include the latest observation results of the observation area. For example, if a change occurs in the observation area between time T2 and time T5, if only the artificial satellite 202 is used, the change will not appear in the observation data until time T5. However, the integrated data is generated as integrated data ID3 and ID4 from the observation data D33 from the artificial satellite 203 at time T3 and the observation data D14 from the artificial satellite 201 at time T4. Since the integrated data ID3 and ID4 include changes that occurred in the observation area between time T2 and time T5, the integrated data ID3 and ID4 also reflect the changes that occurred in the observation area. By analyzing the integrated data based on the observation data observed by multiple satellites, it is possible to grasp changes that occurred within the observation interval of a single satellite in a timely manner. Furthermore, even if an event occurs that prevents the satellite 202 from observing the observation area at time T5, such as bad weather, it is possible to grasp changes in the observation area by analyzing the integrated data generated, including data obtained using the SAR method, which is not affected by weather.

The observation data processing device 101 makes it possible to update data related to the observation area at shorter time intervals, making it possible to apply it to tasks that require real-time analysis. One example of a task is the measurement of NDVI (Normalized Difference Vegetation Index), a vegetation index, using a satellite or the like. While NDVI can be measured more accurately using an optical satellite, optical data is limited to daytime observations and is further restricted by time and weather, such as being unable to observe due to clouds. Therefore, by measuring NDVI using integrated data including SAR data from a SAR satellite, the number of measurements can be increased and real-time performance can be improved.

In addition, because the integrated data is data that combines information from each satellite, for example, when using a learning model to obtain meta-information and analyze an observation area, more frequent and multifaceted analysis is possible than when inputting information from each satellite into the learning model individually.

In addition, it will be possible to use an observation system that currently uses data from a single satellite as a system that can generate integrated data using data from other satellites and enable analysis.

The components of the observation data processing device 101 will now be described. FIG. 3 shows a block diagram of the observation data processing device 101 according to the first embodiment. The observation data processing device 101 has a communication unit 1011, a storage unit 1012, an observation data acquisition unit 1013, an integrated data generation unit 1014, and a meta information acquisition unit 1015. Each component of the observation data processing device 101 can be realized by a processor executing a program stored in a storage device in an information processing device such as a personal computer.

The communication unit 1011 controls communication between the observation data processing device 101 and external information processing devices including the observation data receiving device R. The observation data processing device 101 may also acquire observation data from an information processing device such as a server that records the observation data, and the communication unit 1011 controls communication with other information processing terminals.

The storage unit 1012 stores various information used for processing in the observation data processing device 101. The storage unit 1012 stores a meta information output model 10121.

The meta-information output model 10121 will be described with reference to FIG. 4. The meta-information output model 10121 is a learning model that outputs meta-information of an observation area in response to input of integrated data obtained from observation of the observation area.

The meta-information output model 10121 is trained using the training data LD1 as teacher data. The training data LD1 includes a set of training integrated data based on multiple training observation data obtained by observing the observation area, and meta-information MD0 corresponding to the training integrated data. The training model 10212 is trained using the training integrated data as input and meta-information MD0 as output.

The meta information MD0 has various items depending on the observation target. Meta information for land-based moving objects such as automobiles and animals includes information on the moving object's movement trajectory. This information is obtained from changes in SAR data due to changes in the interference of reflected electromagnetic waves caused by the moving object, and changes in the moving object's position in optical images.

When flooded areas during a disaster are the subject of observation, the meta-information may be information on the extent of the flooded areas. In addition, as meta-information used for managing agricultural crops, NDVI or the like, which indicates the degree of crop growth, may be used. The degree of crop growth is estimated based on a correlation calculated based on the backscattering coefficient of the SAR data and the actually observed degree of crop growth, or on color changes and spectral reflectance characteristics in the optical data. In this case, the meta-information may include information on the actual measured height of the crop.

In detecting buildings, information about new buildings may be used as meta information. Information about new buildings is estimated based on correlations calculated based on the backscattering coefficient of SAR data and information about new buildings that have actually been observed, or color changes in optical data. Information about the type of building may also be obtained based on a map or the like and used as meta information.

When the integrated data generated by the observation data processing device 101 is input to the trained meta information output model 10121, the meta information output model 10121 outputs meta information MD1 corresponding to the integrated data. Note that the meta information output model 10121 does not need to be stored in the observation data processing device 101, and may be stored in another information processing device.

Returning to FIG. 2, the observation data acquisition unit 1013 acquires the observation data observed by the artificial satellite 201 (first observation data) and the observation conditions (first observation conditions) for observing the artificial satellite 201. The observation data acquisition unit 1013 may acquire the observation data and the observation conditions from, for example, the observation data receiving device R, or may acquire the observation data and the observation conditions from a server on which actual observation data and observation data generated by a simulation are recorded. Note that, when generating the integrated data, the observation data processing device 101 may generate the integrated data based on the observation data without acquiring the observation conditions. The same applies to the observation data from the artificial satellite 202 (second observation data), the observation conditions (second observation conditions) for observing the artificial satellite 202, and the observation data and observation conditions from the artificial satellite 203.

The observation conditions include aircraft information, which is information based on the characteristics of the flying object, and environmental information that indicates the observation environment by the flying object. The aircraft information includes the frequency of light used by the flying object for observation, the polarization of light, pulse information, and the beam pattern. In addition, if the flying object forms, for example, a constellation, the constellation identifier and the unit number or identifier of each flying object within the constellation may be included in the aircraft information.

The environmental information includes the orbit direction at the time of observation (northbound orbit ascending or southbound orbit descending), orbit angle, observation direction at the time of observation (left or right), aircraft speed, aircraft attitude information, and parameters of the receiving system. If the observation data is image data, the environmental information may also include information on the resolution of the image. If the observation data is SAR image data, the environmental information may also include information on the angle of incidence of the irradiation signal for each pixel of the image. If noise removal is performed on the observation data, the environmental information may also include noise information, the type of noise filter, and parameters for noise removal.

The integrated data generation unit 1014 performs a process of generating integrated data. For example, based on observation data D11, which is information on the observation area at time T1 and observed by satellite 201, and observation data D22, which is information on the observation area at time T2 and observed by satellite 202, the integrated data generation unit 1014 generates integrated data ID1 based on first information based on observation data D11 and second information based on observation data D22.

The meta information acquisition unit 1015 inputs the integrated data into the meta information output model 10121 and acquires meta information of the observation area. The acquired meta information may be presented to a user of the observation data processing device 101 by a display unit of the observation data processing device 101, or may be transmitted from the observation data processing device 101 to another information processing device and presented to the user.

The integrated data generation process will be described with reference to FIG. 5. In step S501, the observation data acquisition unit 1013 acquires first observation data at a first timing. For example, the observation data acquisition unit 1013 acquires observation data D11 at time T1.

In step S502, the observation data acquisition unit 1013 acquires second observation data at a second timing. For example, the observation data acquisition unit 1013 acquires observation data D22 at time T2.

In step S503, the integrated data generation unit 1014 acquires first information based on the first observation data. Here, the first information is, for example, image data generated based on the first observation data, or meta information obtained from a signal or image data from the artificial satellite 201. In other words, the data included in the integrated data may be information based on the observation data. Furthermore, the first information may be, for example, the observation data itself from the artificial satellite 201.

In step S504, the integrated data generation unit 1014 acquires second information based on the second observation data. The second information is similar to the first information.

In step S505, the integrated data generation unit 1014 generates integrated data including the first information and the second information. At this time, the integrated data generation unit 1014 may store the generated integrated data in the storage unit 1012. In addition, the integrated data generation unit 1014 may transmit the generated integrated data to an external information processing device.

The integrated data generation unit 1014 may generate integrated data each time observation data is acquired. For example, the integrated data generation unit 1014 may generate integrated data such that observation data D14 obtained by the satellite 201 observing the observation area at time T4 is replaced with observation data D11 obtained at time T1 before time T4.

The meta-information acquisition process based on the integrated data will be described with reference to FIG. 6. In step S601, the meta-information acquisition unit 1015 acquires the integrated data generated by the integrated data generation unit 1014. The meta-information acquisition unit 1015 acquires the integrated data, for example, by referring to the integrated data stored in the storage unit 1012. In step S602, the meta-information acquisition unit 1015 inputs the integrated data to the meta-information output model 10121. In step S603, the meta-information acquisition unit 1015 acquires meta-information of the observation area from the meta-information output model 10121.

By carrying out the above processing, the observation data processing device 101 can generate data obtained by observing an observation area using multiple flying objects at a high frequency, and obtain meta information about the observation area.

Second Embodiment
A second embodiment will be described. In the second and subsequent embodiments, a description of matters common to the first embodiment will be omitted, and only the differences will be described. The second embodiment differs from the first embodiment in that observed data is converted using a conversion model, and integrated data is generated based on the converted data.

Figure 7 shows a block diagram of an observation data processing device 101A according to the second embodiment. In addition to the components of the observation data processing device 101, the observation data processing device 101A has a memory unit 1012A having a first conversion model 10122 and a second conversion model 10123, and a converted observation data acquisition unit 1016.

The first conversion model 10122 will be described with reference to FIG. 8. The first conversion model 10122 is a learning model that outputs converted observation data in response to input of third observation data obtained by observing the observation region, third observation conditions, and third observation timing.

The first conversion model 10122 is trained using the training data LD2 as teacher data. The training data LD2 includes sets of training observation data obtained by observing the training observation area, training observation conditions, and training observation timing, and the corresponding converted observation data. The converted observation data is data observed under the first observation conditions, and the observation timing is the training observation timing. The first conversion model 10122 is trained using the training observation data, training observation conditions, and training observation timing as input, and the converted observation data as output.

The first conversion model 10122 is a model that generates observation data obtained by changing the observation conditions of observation data at a certain timing. The first conversion model 10122 converts observation data with different observation conditions into observation data under certain observation conditions (e.g., first observation conditions), and generates integrated data.

The converted observation data acquisition unit 1016 acquires converted observation data in which the observation data is converted using the first conversion model 10122 and the second conversion model 10123 described below.

The process of generating integrated data using the first conversion model 10122 in the observation data processing device 101A will be described with reference to Figures 9 and 10. Here, it is assumed that the satellites 201, 202, and 203 acquire observation data based on a common observation method (e.g., SAR method). The observation system 10 further includes a satellite 204 that acquires observation data based on an observation method (e.g., optical method) different from that of the satellites 201, 202, and 203.

In step S901, the observation data acquisition unit 1013 acquires first observation data at a first timing. For example, the observation data acquisition unit 1013 acquires observation data D11 at time T1.

In step S902, the observation data acquisition unit 1013 acquires second observation data at a second timing. For example, the observation data acquisition unit 1013 acquires observation data D42 at time T2.

In step S903, the observation data acquisition unit 1013 acquires third observation data at a third timing. For example, the observation data acquisition unit 1013 acquires observation data D23 at time T3.

In step S904, the converted observation data acquisition unit 1016 acquires third observation conditions for the third observation data. For example, the converted observation data acquisition unit 1016 acquires observation conditions for the observation data D23. More specifically, the converted observation data acquisition unit 1016 acquires aircraft information of the artificial satellite 202 and environmental information of the observation environment of the observation data D23 as the observation conditions.

In step S905, the converted observation data acquisition unit 1016 inputs the third observation data, the third observation condition, and the third timing to the first conversion model 10122. For example, the converted observation data acquisition unit 1016 inputs the observation data D23, the observation condition for the observation data D23, and the time T3 to the first conversion model 10122.

In step S906, the converted observation data acquisition unit 1016 acquires the converted observation data into which the third observation data has been converted. For example, the converted observation data acquisition unit 1016 acquires the converted observation data D13 into which the observation data D23 has been converted from the first conversion model 10122.

In step S907, the integrated data generation unit 1014 updates the first information with the first information based on the converted observation data to generate integrated data including the first information and the second information. For example, the integrated data generation unit 1014 generates integrated data ID2 in which the observation data D11 and the observation data D42 are integrated. Next, the integrated data generation unit 1014 generates integrated data ID3 in which the observation data D11 is updated with the converted observation data D13.

Similarly, the integrated data generation unit 1014 generates integrated data ID4 based on converted observation data D14 into which observation data D34 has been converted. The integrated data generation unit 1014 also generates integrated data ID5 in which the converted observation data D14 has been updated by observation data D15 in the integrated data.

As a result, even if there are differences in aircraft characteristics or observation environments between satellite 201 and satellites 202 and 203, for example, it is possible to align the observation data used to generate the integrated data as data observed under the observation conditions of satellite 201. As a result, integrated data is generated in which differences between the observation means are reduced. It also becomes possible to generate data observed under the observation conditions of satellite 201 frequently, and as a result, it becomes possible to update the integrated data frequently. By using such integrated data, meta-information of the observation area can be obtained with greater accuracy.

The second conversion model 10123 will be described with reference to FIG. 11. The second conversion model 10123 is a learning model that outputs converted observation data in response to input of first observation data obtained by observing the observation region, first observation conditions, and first observation timing.

The second conversion model 10123 is trained using the training data LD3 as teacher data. The training data LD3 includes sets of training observation data, training observation conditions, and training observation timing obtained by observing the training observation area, and the corresponding converted observation data. The converted observation data is data observed under reference observation conditions, and the observation timing is the training observation timing. The learning model 10212 is trained using the training observation data, training observation conditions, and training observation timing as inputs and the converted observation data as output. Here, the reference observation conditions are predetermined observation conditions that are different from the observation conditions of each of the artificial satellites 201, 202, and 203. The reference observation conditions are conditions that are predetermined by the user.

The second conversion model 10123 is a model that generates observation data obtained by changing the observation conditions of observation data at a certain timing. The second conversion model 10123 converts observation data obtained under different observation conditions into observation data obtained under standard observation conditions, and generates integrated data.

The process of generating integrated data using the second conversion model 10123 in the observation data processing device 101A will be described with reference to Figures 12 and 13. Here, it is assumed that the satellites 201, 202, and 203 acquire observation data based on a common observation method (e.g., SAR method). The observation system 10 further includes a satellite 204 that acquires observation data based on an observation method (e.g., optical method) different from that of the satellites 201, 202, and 203.

In step S1201, the observation data acquisition unit 1013 acquires first observation data at a first timing. For example, the observation data acquisition unit 1013 acquires observation data D11 at time T1.

In step S1202, the converted observation data acquisition unit 1016 acquires the first observation conditions for the first observation data. For example, the converted observation data acquisition unit 1016 acquires the observation conditions for the observation data D11. More specifically, the converted observation data acquisition unit 1016 acquires the aircraft information of the artificial satellite 201 and the environmental information of the observation environment as the observation conditions.

In step S1203, the converted observation data acquisition unit 1016 inputs the first observation data, the first observation condition, and the first timing to the second conversion model 10123. For example, the converted observation data acquisition unit 1016 inputs the observation data D11, the observation condition for the observation data D11, and the time T1 to the second conversion model 10123.

In step S1204, the converted observation data acquisition unit 1016 acquires the first converted observation data into which the first observation data is converted. For example, the converted observation data acquisition unit 1016 acquires the converted observation data D51 into which the observation data D11 is converted from the first conversion model 10122.

In step S1205, the observation data acquisition unit 1013 acquires second observation data at a second timing. For example, the observation data acquisition unit 1013 acquires observation data D42 at time T2.

In step S1206, the integrated data generation unit 1014 generates integrated data based on the first information based on the first converted observation data and the second information based on the second observation data. For example, the integrated data generation unit 1014 generates integrated data ID2 based on the converted observation data D51 and the observation data D42.

In step S1207, the observation data acquisition unit 1013 acquires third observation data at a third timing. For example, the observation data acquisition unit 1013 acquires observation data D23 at time T3.

In step S1208, the converted observation data acquisition unit 1016 acquires third observation conditions for the third observation data. For example, the converted observation data acquisition unit 1016 acquires observation conditions for the observation data D23. More specifically, the converted observation data acquisition unit 1016 acquires aircraft information about the artificial satellite 201 and environmental information about the observation environment as the observation conditions.

In step S1209, the converted observation data acquisition unit 1016 inputs the third observation data, the third observation condition, and the third timing to the second conversion model 10123. For example, the converted observation data acquisition unit 1016 inputs the observation data D23, the observation condition for the observation data D23, and the time T3 to the second conversion model 10123.

In step S1210, the converted observation data acquisition unit 1016 acquires the third converted observation data into which the third observation data is converted. For example, the converted observation data acquisition unit 1016 acquires the converted observation data D53 into which the observation data D23 is converted from the first conversion model 10122.

In step S1211, the integrated data generation unit 1014 updates the first information based on the first converted observation data with the first information based on the second converted observation data to generate integrated data including the first information and the second information. For example, the integrated data generation unit 1014 generates integrated data ID3 in which the observation data D51 is updated by the converted observation data D53.

Similarly, the integrated data generation unit 1014 generates integrated data ID4 based on converted observation data D54 into which observation data D34 is converted. Also, the integrated data generation unit 1014 generates integrated data ID5 based on converted observation data D55 into which observation data D15 is converted.

As a result, even if there are differences in aircraft characteristics or observation environments between satellite 201 and satellites 202 and 203, for example, it is possible to align the observation data used to generate the integrated data as data observed under specified observation conditions. This allows integrated data to be generated in which differences between the observation means are reduced. By using such integrated data, meta-information of the observation area can be obtained with greater accuracy.

Third Embodiment
A third embodiment will be described. The third embodiment differs from the first and second embodiments in that observation data is predicted using a prediction model, and integrated data is generated based on the predicted data. Here, prediction of observation data means prediction of observation data observed at a certain timing from other observation data.

Figure 14 shows a block diagram of an observation data processing device 101B according to the third embodiment. In addition to the components of the observation data processing device 101, the observation data processing device 101B has a memory unit 1012B having a first estimation model 10124, a second estimation model 10125, and a third estimation model 10126, and an estimated observation data acquisition unit 1017. The estimated observation data acquisition unit 1017 acquires estimated observation data in which observation data is estimated using the first estimation model 10124, the second estimation model 10125, or the third estimation model 10126 described below.

The first estimation model 10124 will be described with reference to FIG. 15. The first estimation model 10124 is a learning model that outputs estimated observation data in response to input of first observation data obtained by observing the observation area and a second observation timing. The input first observation data is data observed based on a first observation method (e.g., a SAR method), and the output estimated observation data is data observed based on a second observation method (e.g., an optical method) that is different from the first observation method.

The first estimation model 10124 is trained using the training data LD4 as teacher data. The training data LD4 includes a set of training observation data and training observation timing A obtained by observing the training observation area, and the corresponding estimated observation data. The estimated observation data is data observed based on the first observation method, and its observation timing is timing A. The first estimation model 10124 is trained using the training observation data and training observation timing as inputs, and the estimated observation data as output.

The first estimation model 10124 is a model that generates observation data obtained by changing the observation method of observation data at a certain timing to another observation method. The first estimation model 10124 estimates observation data obtained using a different observation method based on certain observation data, and generates integrated data.

The process of generating integrated data using the first estimation model 10124 in the observation data processing device 101B will be described with reference to Figures 16 and 17. Here, it is assumed that the artificial satellite 201 acquires observation data based on a first observation method (e.g., a SAR method), the artificial satellites 202 and 204 acquire observation data based on a second observation method (e.g., an optical method), and the artificial satellite 203 acquires observation data based on a third observation method (e.g., an observation method using meteorological data).

In step S1601, the observation data acquisition unit 1013 acquires first observation data at a first timing. For example, the observation data acquisition unit 1013 acquires observation data D11 at time T1.

In step S1602, the observation data acquisition unit 1013 acquires second observation data at a second timing. For example, the observation data acquisition unit 1013 acquires observation data D42 at time T2.

In step S1603, the observation data acquisition unit 1013 acquires third observation data at a third timing. For example, the observation data acquisition unit 1013 acquires observation data D23 at time T3.

In step S1604, the converted observation data acquisition unit 1016 acquires third observation conditions for the third observation data. For example, the converted observation data acquisition unit 1016 acquires observation conditions for the observation data D23. More specifically, the converted observation data acquisition unit 1016 acquires aircraft information about the artificial satellite 202 and environmental information about the observation environment of the observation data D23 as the observation conditions.

In step S1605, the converted observation data acquisition unit 1016 inputs the third observation data, the third observation condition, and the third timing to the first estimation model 10124. For example, the converted observation data acquisition unit 1016 inputs the observation data D23, the observation condition for the observation data D23, and the time T3 to the first conversion model 10122.

In step S1606, the converted observation data acquisition unit 1016 acquires the converted observation data into which the third observation data has been converted. For example, the converted observation data acquisition unit 1016 acquires the estimated observation data D13 into which the observation data D23 has been converted from the first estimated model 10124.

In step S1607, the integrated data generation unit 1014 updates the first information based on the estimated observation data to generate integrated data including the first information and the second information. For example, the integrated data generation unit 1014 generates integrated data ID2 in which the observed data D11 and the observed data D42 are integrated. Next, the integrated data generation unit 1014 generates integrated data ID3 in which the observed data D11 is updated by the estimated observation data D13.

Similarly, the integrated data generation unit 1014 generates integrated data ID4 based on converted observation data D14 into which observation data D34 has been converted. The integrated data generation unit 1014 also generates integrated data ID5 in which the converted observation data D14 has been updated by observation data D15 in the integrated data.

As a result, even if there are differences in the observation methods between satellites 201, 202, and 203, for example, it is possible to align the observation data used to generate the integrated data as data observed under a certain observation method, for example, the observation method of satellite 201. It is also possible to generate data observed under the observation method of satellite 201 with high frequency. As a result, the differences between the observation methods are balanced, and integrated data is generated with high frequency. By using such integrated data, meta-information of the observation area can be obtained with higher accuracy.

The second estimation model 10125 will be described with reference to FIG. 18. The second estimation model 10125 is a learning model that outputs estimated observation data in response to input of first observation data obtained by observing the observation area, a first observation timing, and a second observation timing. The input first observation data is data observed based on a first observation method (e.g., a SAR method), and the output estimated observation data is data observed based on a second observation method (e.g., an optical method) that is different from the first observation method.

The second estimation model 10125 is trained using the training data LD5 as teacher data. The training data LD5 includes a set of training observation data obtained by observing the training observation area, a training observation timing A, and a timing B different from the training observation timing A, and the corresponding estimated observation data. The estimated observation data is data observed based on the first observation method, and its observation timing is timing B. The second estimation model 10125 is trained using the training observation data, the training observation timing, and timing B as inputs, and the estimated observation data as output.

The second estimation model 10125 is a model that changes the observation method of observation data at a certain timing to another observation method, and generates observation data obtained at another timing. The second estimation model 10125 estimates observation data with different observation methods and observation timing based on certain observation data, and generates integrated data.

The process of generating integrated data using the second estimation model 10125 in the observation data processing device 101B will be described with reference to Figures 19 and 22. Here, it is assumed that the satellites 201 and 203 acquire observation data based on a first observation method (e.g., SAR method), and the satellite 202 acquires observation data based on a second observation method (e.g., optical method).

In step S1901, the observation data acquisition unit 1013 acquires first observation data at a first timing. For example, the observation data acquisition unit 1013 acquires observation data D22 at time T2.

In step S1902, the inferred observation data acquisition unit 1017 inputs the first observation data, the first timing, and the second timing to the first estimation model. For example, as shown in FIG. 22, the inferred observation data acquisition unit 1017 inputs the observation data D22, time T2, and time T3 to the second estimation model 10125.

In step S1903, the inferred observation data acquisition unit 1017 acquires inferred observation data at the second timing from the second inference model 10125. For example, as shown in FIG. 22, the inferred observation data acquisition unit 1017 acquires inferred observation data D63 at time T3 from the second inference model 10125.

In step S1904, the integrated data generation unit 1014 generates integrated data based on the first information based on the first observation data and the second information based on the inferred observation data. For example, as shown in FIG. 22, the integrated data generation unit 1014 generates integrated data ID3 based on the observation data D11 and the inferred observation data D63.

As a result, even if there is a difference in the observation method between satellite 202 and satellites 201 and 203, for example, it is possible to align the observation data used to generate the integrated data as data observed under the observation methods of satellites 201 and 203. As a result, integrated data is generated in which the differences between the observation methods are reduced. By using such integrated data, meta-information of the observation area can be obtained with greater accuracy.

The third estimation model 10126 will be described with reference to FIG. 20. The third estimation model 10126 is a learning model that outputs estimated observation data in response to input of first observation data obtained by observing the observation area, a third observation timing prior to the first observation timing, and the first observation timing. The input first observation data and the output estimated observation data are data observed based on a common observation method (e.g., SAR method). The third estimation model 10126 is a model that performs time series regression prediction and generates estimated observation data.

The third estimation model 10126 is trained using the training data LD6 as teacher data. The training data LD6 includes sets of training observation data obtained by observing the training observation area, training observation timing A, and timing C prior to training observation timing A, and the corresponding estimated observation data. The training observation data and estimated observation data are data observed based on the first observation method, and the observation timing is timing C. The third estimation model 10126 is trained using the training observation data, training observation timing, and timing C as inputs, and the estimated observation data as output.

The third estimation model 10126 is a model that generates observation data obtained at a timing after a certain timing based on observation data at that timing, without changing the observation method. The third estimation model 10126 estimates observation data observed at different observation timings based on the certain observation data, and generates integrated data.

The process of generating integrated data using the third estimation model 10126 in the observation data processing device 101B will be described with reference to Figures 20 and 21. Here, it is assumed that the satellites 201 and 203 acquire observation data based on a first observation method (e.g., SAR method), and the satellite 202 acquires observation data based on a second observation method (e.g., optical method).

In step S2101, the observation data acquisition unit 1013 acquires third observation data when the observation area is observed at a third timing prior to the first timing. For example, the observation data acquisition unit 1013 acquires observation data D22 when the observation area is observed at time T2 prior to time T5.

In step S2102, the inferred observation data acquisition unit 1017 inputs the third observation data, the third timing, and the first timing to the inferred model. For example, as shown in FIG. 22, the inferred observation data acquisition unit 1017 inputs the observation data D22, time T2, and time T5 to the third inferred model 10126.

In step S2103, the inferred observation data acquisition unit 1017 acquires inferred observation data at the first timing from the third inference model 10126. For example, as shown in FIG. 22, the inferred observation data acquisition unit 1017 acquires inferred observation data D75 at time T5 from the third inference model 10126.

In step S2104, the integrated data generation unit 1014 generates integrated data based on the first information based on the estimated observation data at the first timing and the second information based on the second observation data. For example, as shown in FIG. 22, the integrated data generation unit 1014 generates integrated data ID5 based on the estimated observation data D53 and the observation data D14.

As a result, even if, for example, the satellite 202 is unable to perform observation at a certain point in time, it is possible to estimate the observation data used to generate the integrated data and use it as the integrated data. As a result, even if some of the observation data cannot be obtained, integrated data with short time intervals can be generated. By using such integrated data, meta-information of the observation area can be obtained with greater accuracy.

The above-described embodiments are intended to facilitate understanding of the present invention, and are not intended to limit the present invention. The elements of the embodiments, as well as their arrangement, conditions, shape, size, etc., are not limited to those exemplified, and may be modified as appropriate. In addition, configurations shown in different embodiments may be partially substituted or combined.

10...observation system, 101, 101A, 101B...observation data processing device, 201, 202, 203...artificial satellite, 1013...observation data acquisition unit, 1014...integrated data generation unit, 1015...meta information acquisition unit, 1016...converted observation data acquisition unit, 1017...estimated observation data acquisition unit

Claims (16)

On the computer,
acquiring first observation data, which is information on an observation area at a first timing and is observed under first observation conditions by a first flying object, and the first observation conditions;
acquiring second observation data, which is information on the observation area at a second timing different from the first timing, observed under second observation conditions by a second flying object different from the first flying object, and observed based on an observation method common to the first observation data;
inputting the second observation data, the second observation conditions, and the second timing into a conversion model which receives learning observation data, which is information about a learning observation area, learning observation conditions corresponding to the learning observation data, and a learning observation timing at which the learning observation data was observed, and outputs converted observation data into which the learning observation data is converted, the converted observation data indicating data observed in the learning observation area under the first observation conditions at the learning observation timing, and acquiring the converted observation data into which the second observation data is converted;
generating integrated data including first information based on the first observation data and second information based on the converted observation data. On the computer,
acquiring first observation data, which is information on an observation area at a first timing and is observed under first observation conditions by a first flying object, and the first observation conditions;
acquiring second observation data, which is information on the observation area at a second timing different from the first timing, observed under second observation conditions by a second flying object different from the first flying object, and observed based on an observation method common to the first observation data;
a conversion model that receives input of learning observation data, which is information on a learning observation area, learning observation conditions corresponding to the learning observation data, and a learning observation timing at which the learning observation data was observed, and outputs converted observation data obtained by converting the learning observation data, the converted observation data indicating data observed in the observation area under reference observation conditions at the learning observation timing, inputting the first observation data, the first observation conditions, and the first timing, and acquiring first converted observation data as the converted observation data obtained by converting the first observation data;
inputting the second observation data, the second observation condition, and the second timing into the conversion model, and acquiring second converted observation data as the converted observation data obtained by converting the second observation data at the second timing;
generating integrated data including first information based on the first converted observation data and second information based on the second converted observation data. 3. The program according to claim 1 or 2 ,
the first observation condition includes first aircraft information of the first flying object and first environmental information indicating an observation environment by the first flying object;
The second observation condition includes second aircraft information of the second flying object and second environmental information indicating an observation environment by the second flying object . On the computer,
acquiring first observation data, which is information on an observation area at a first timing and is observed by a first flying object based on a first observation method;
acquiring second observation data, which is information on the observation area at a second timing different from the first timing, observed by a second flying object different from the first flying object based on a second observation method different from the first observation method;
inputting the second observation data and the second timing into an estimation model which receives learning observation data and a learning observation timing at which the learning observation data was observed, and outputs estimated observation data indicating data at which the observation domain was observed based on the first observation method at the learning observation timing, and acquiring the estimated observation data at the second timing;
generating integrated data including first information based on the first observation data and second information based on the estimated observation data. On the computer,
acquiring first observation data, which is information on an observation area at a first timing and is observed by a first flying object based on a first observation method;
inputting the first observation data, the first timing, and a second timing different from the first timing into an estimation model which receives as input learning observation data, a learning observation timing at which the learning observation data was observed, and a timing different from the learning observation timing, and outputs estimated observation data indicating data obtained by observing the observation area based on a second observation method different from the first observation method at the different timing, the first observation data, the first timing, and a second timing different from the first timing, and acquiring the estimated observation data at the second timing;
generating integrated data including first information based on the first observation data and second information based on the estimated observation data. On the computer,
generating integrated data including first information based on the first observation data and second information based on the second observation data, the first information being information on an observation area at a first timing, the first observation data being observed by a first flying object based on a first observation method, and second observation data being information on the observation area at a second timing different from the first timing, the second observation data being observed by a second flying object different from the first flying object based on a second observation method different from the first observation method;
acquiring third observation data when the observation area is observed at a third timing that is earlier than the first timing;
inputting the third observation data, the third timing, and the first timing into an estimation model which receives learning observation data, a learning observation timing at which the learning observation data was observed, and a prediction timing after the learning observation timing, and outputs estimated observation data indicating data at which the observation region was observed based on the first observation method at the prediction timing, and acquiring the estimated observation data at the first timing;
generating the integrated data,
generating the integrated data including the first information based on the estimated observation data at the first timing. an observation data acquisition unit that acquires first observation data, which is information about an observation area at a first timing, observed by a first flying object under first observation conditions, and the first observation conditions, and acquires second observation data, which is information about the observation area at a second timing different from the first timing, observed by a second flying object different from the first flying object under second observation conditions, and observed based on an observation method common to the first observation data;
a converted observation data acquisition unit which inputs the second observation data, the second observation conditions and the second timing into a conversion model which receives input of learning observation data, which is information about a learning observation area, learning observation conditions corresponding to the learning observation data, and a learning observation timing at which the learning observation data was observed, and outputs converted observation data obtained by converting the learning observation data, the converted observation data indicating data observed in the learning observation area under the first observation conditions at the learning observation timing, and acquires the converted observation data obtained by converting the second observation data;
an integrated data generation unit that generates integrated data including first information based on the first observation data and second information based on the converted observation data. an observation data acquisition unit that acquires first observation data, which is information about an observation area at a first timing, observed by a first flying object under first observation conditions, and the first observation conditions, and acquires second observation data, which is information about the observation area at a second timing different from the first timing, observed by a second flying object different from the first flying object under second observation conditions, and observed based on an observation method common to the first observation data;
a converted observation data acquisition unit that inputs the first observation data, the first observation condition, and the first timing into a conversion model that receives input of learning observation data, which is information about a learning observation area, learning observation conditions corresponding to the learning observation data, and a learning observation timing at which the learning observation data was observed, and outputs converted observation data obtained by converting the learning observation data, the converted observation data indicating data observed in the observation area under reference observation conditions at the learning observation timing, and acquires first converted observation data as the converted observation data obtained by converting the first observation data, and inputs the second observation data, the second observation condition, and the second timing into the conversion model, and acquires second converted observation data as the converted observation data obtained by converting the second observation data at the second timing;
an integrated data generation unit that generates integrated data including first information based on the first converted observation data and second information based on the second converted observation data. an observation data acquisition unit that acquires first observation data, which is information on an observation area at a first timing, observed by a first flying object based on a first observation method, and acquires second observation data, which is information on the observation area at a second timing different from the first timing, observed by a second flying object different from the first flying object based on a second observation method different from the first observation method;
an inferred observation data acquisition unit that inputs the second observation data and the second timing to an inference model that receives learning observation data and a learning observation timing at which the learning observation data was observed, and outputs inferred observation data indicating data at which the observation domain was observed based on the first observation method at the learning observation timing, and acquires the inferred observation data at the second timing;
an integrated data generation unit that generates integrated data including first information based on the first observation data and second information based on the estimated observation data. an observation data acquisition unit that acquires first observation data, which is information about an observation area at a first timing and is observed by a first flying object based on a first observation method;
an inferred observation data acquisition unit that inputs learning observation data, a learning observation timing at which the learning observation data was observed, and a timing different from the learning observation timing, and outputs inferred observation data indicating data observed in the observation area based on a second observation method different from the first observation method at the different timing, and inputs the first observation data, the first timing, and a second timing different from the first timing to an inferred observation data acquisition unit that acquires the inferred observation data at the second timing;
an integrated data generation unit that generates integrated data including first information based on the first observation data and second information based on the estimated observation data. an integrated data generating unit that generates integrated data including first information based on the first observation data and second information based on the second observation data, the first information being information on an observation area at a first timing, the first observation data being observed by a first flying object based on a first observation method, and second observation data being information on the observation area at a second timing different from the first timing, the second observation data being observed by a second flying object different from the first flying object based on a second observation method different from the first observation method;
an observation data acquisition unit that acquires third observation data when the observation area is observed at a third timing that is earlier than the first timing;
an estimated observation data acquisition unit that inputs the third observation data, the third timing, and the first timing to an estimation model that receives learning observation data, a learning observation timing at which the learning observation data was observed, and a prediction timing after the learning observation timing, and outputs estimated observation data indicating data at which the observation area was observed based on the first observation method at the prediction timing, and acquires the estimated observation data at the first timing;
The integrated data generation unit
An information processing device that generates the integrated data including the first information based on the estimated observation data at the first timing. The computer
acquiring first observation data, which is information on an observation area at a first timing and is observed under first observation conditions by a first flying object, and the first observation conditions;
acquiring second observation data, which is information on the observation area at a second timing different from the first timing, observed under second observation conditions by a second flying object different from the first flying object, and observed based on an observation method common to the first observation data;
inputting the second observation data, the second observation conditions, and the second timing into a conversion model which receives learning observation data, which is information about a learning observation area, learning observation conditions corresponding to the learning observation data, and a learning observation timing at which the learning observation data was observed, and outputs converted observation data into which the learning observation data is converted, the converted observation data indicating data observed in the learning observation area under the first observation conditions at the learning observation timing, and acquiring the converted observation data into which the second observation data is converted;
generating integrated data including first information based on the first observation data and second information based on the converted observation data. The computer
acquiring first observation data, which is information on an observation area at a first timing and is observed under first observation conditions by a first flying object, and the first observation conditions;
acquiring second observation data, which is information on the observation area at a second timing different from the first timing, observed under second observation conditions by a second flying object different from the first flying object, and observed based on an observation method common to the first observation data;
a conversion model that receives input of learning observation data, which is information on a learning observation area, learning observation conditions corresponding to the learning observation data, and a learning observation timing at which the learning observation data was observed, and outputs converted observation data obtained by converting the learning observation data, the converted observation data indicating data observed in the observation area under reference observation conditions at the learning observation timing, inputting the first observation data, the first observation conditions, and the first timing, and acquiring first converted observation data as the converted observation data obtained by converting the first observation data;
inputting the second observation data, the second observation condition, and the second timing into the conversion model, and acquiring second converted observation data as the converted observation data obtained by converting the second observation data at the second timing;
generating integrated data including first information based on the first converted observation data and second information based on the second converted observation data. The computer
acquiring first observation data, which is information on an observation area at a first timing and is observed by a first flying object based on a first observation method;
acquiring second observation data, which is information on the observation area at a second timing different from the first timing, observed by a second flying object different from the first flying object based on a second observation method different from the first observation method;
inputting the second observation data and the second timing into an estimation model which receives learning observation data and a learning observation timing at which the learning observation data was observed, and outputs estimated observation data indicating data at which the observation domain was observed based on the first observation method at the learning observation timing, and acquiring the estimated observation data at the second timing;
generating integrated data including first information based on the first observation data and second information based on the inferred observation data. The computer
acquiring first observation data, which is information on an observation area at a first timing and is observed by a first flying object based on a first observation method;
inputting the first observation data, the first timing, and a second timing different from the first timing into an estimation model which receives as input learning observation data, a learning observation timing at which the learning observation data was observed, and a timing different from the learning observation timing, and outputs estimated observation data indicating data obtained by observing the observation area based on a second observation method different from the first observation method at the different timing, the first observation data, the first timing, and a second timing different from the first timing, and acquiring the estimated observation data at the second timing;
generating integrated data including first information based on the first observation data and second information based on the inferred observation data. The computer
generating integrated data including first information based on the first observation data and second information based on the second observation data, the first information being information on an observation area at a first timing, the first observation data being observed by a first flying object based on a first observation method, and second observation data being information on the observation area at a second timing different from the first timing, the second observation data being observed by a second flying object different from the first flying object based on a second observation method different from the first observation method;
acquiring third observation data when the observation area is observed at a third timing that is earlier than the first timing;
inputting the third observation data, the third timing, and the first timing into an estimation model which receives learning observation data, a learning observation timing at which the learning observation data was observed, and a prediction timing after the learning observation timing, and outputs estimated observation data indicating data at which the observation domain was observed based on the first observation method at the prediction timing, and acquiring the estimated observation data at the first timing;
generating the integrated data,
An information processing method comprising: generating the integrated data including the first information based on the estimated observation data at the first timing.

Images