Information processing device, information processing method, and computer program

Abstract

A device according to one aspect of the present disclosure is an information processing device comprising a control unit that, by grouping a plurality of similar pixels in an SAR image including a set of a plurality of vehicles positioned adjacent to each other into irregular small regions, generates a divided image in which the plurality of vehicles are divided into different small regions.

Description

Information processing device, information processing method, and computer program

[0001] This application claims priority to Japanese Patent Application No. 2024-122130, filed July 29, 2024, and incorporates by reference all of the contents of that application.

[0002] Optical images acquired from optical cameras on artificial satellites and SAR images acquired by SAR (Synthetic Aperture Radar) mounted on artificial satellites are sometimes used to understand traffic conditions, etc. Of these, optical images are images acquired from optical satellites based on the reflection of sunlight on the Earth's surface, and therefore have the advantage of allowing for more visual observation of the state of the ground, but they make it difficult to observe the state of the ground surface at night or on cloudy days when sunlight does not reach the Earth's surface.

[0003] SAR images are obtained by transmitting radio waves from a SAR mounted on a satellite toward the Earth's surface and receiving the radio waves reflected by the surface with a sensor on the satellite. This allows for better observation of the Earth's surface conditions than optical satellites, even at night or on cloudy days.

[0004] Patent Document 1 describes a traffic control system that uses an SAR mounted on a satellite to capture images of a target area including roads, obtains image data, and grasps the traffic conditions within the target area based on the image data.

[0005] Specifically, the traffic control system of Patent Document 1 extracts driving area data corresponding to a driving area such as a road based on map data such as a road map and the image data, and determines the number of vehicles in the driving area data based on radar reflection waves from vehicles on the road.Similarly, the traffic control system of Patent Document 1 determines the number of vehicles parked in a parking lot based on radar reflection waves from parking areas.

[0006] Japanese Patent Application Laid-Open No. 2001-101572

[0007] An apparatus according to one aspect of the present disclosure is an information processing apparatus that includes a control unit that generates a divided image in which a set of multiple vehicles located close to each other is divided into separate small regions by grouping similar pixels from an SAR image that includes the set of multiple vehicles located close to each other into small, irregular regions.

[0008] The embodiments of the present disclosure may be realized by an apparatus, a system, a method, an integrated circuit, a computer program, or a non-transitory computer-readable recording medium, or any combination thereof. The recording medium may be either volatile or non-volatile. The apparatus may be composed of multiple individual devices. When composed of multiple individual devices, they may be arranged in a single housing or may be arranged separately in two or more separate housings.

[0009] FIG. 1 is a schematic diagram showing an example of the configuration of an information providing system. FIG. 2 is a block diagram showing an example of the configuration of an information providing server and an in-vehicle device. FIG. 3 is a block diagram showing an example of the internal configuration of the information providing server. FIG. 4 is a flowchart illustrating a series of processes executed by a control unit. FIG. 5 is a flowchart showing detailed steps of a segmented image generation process. FIG. 6 is an example of an SAR image used in the segmented image generation process. FIG. 7 is a schematic diagram showing how segmented images are generated from an SAR image. FIG. 8 is a flowchart showing an example of a traffic condition detection process. FIG. 9 is a flowchart showing another example of a traffic condition detection process. FIG. 10 is an explanatory diagram showing an example of a road map screen display including events provided by the information providing server.

[0010] <Problem to be Solved by the Present Disclosure> Because SAR images are information that express the bounce strength of radio waves as brightness values ​​(grayscale information), it is difficult to recognize what objects are included in the images compared to optical images that contain information about the color shades of objects. For example, even if an optical image allows one to recognize whether an object included in the image is a building or a vehicle based on color information such as RGB, in SAR images, the object is expressed using grayscale brightness values, so it may not be easy to recognize the object.

[0011] In this regard, Patent Document 1 discloses that the number of vehicles on a road or in a parking lot can be determined based on image data acquired by SAR, but does not disclose a specific method for determining the number of vehicles.

[0012] In view of the above-described conventional problems, the present disclosure aims to provide an information processing device and the like that can more appropriately recognize vehicles based on SAR images.

[0013] Effect of the Present Disclosure According to the present disclosure, vehicles can be more appropriately recognized based on SAR images.

[0014] <Outline of Embodiments of the Present Disclosure> Below, an outline of embodiments of the present disclosure will be listed and described.

[0015] (1) The device according to this embodiment is an information processing device that includes a control unit that generates a segmented image in which a set of multiple vehicles located close to each other is divided into small, irregularly shaped regions by grouping similar pixels from a SAR image that includes the set of multiple vehicles located close to each other.

[0016] According to the information processing device of this embodiment, by generating a segmented image based on a SAR image, it is possible to distinguish between vehicles included in a group and other objects, thereby enabling more appropriate vehicle recognition.

[0017] (2) In the information processing device of (1) above, the control unit may group a plurality of pixels in the SAR image that are close in distance into the small region.

[0018] This allows the similarity between multiple pixels to be determined based on the distance between the pixels.

[0019] (3) In the information processing device of (2) above, the control unit may group a plurality of pixels in the SAR image that have similar distances and luminance values ​​into the small region.

[0020] This allows the similarity between multiple pixels to be determined based on the pixel distance and brightness value.

[0021] (4) In the information processing device according to any one of (1) to (3), the control unit may generate the segmented image based on the SAR image that has been subjected to threshold processing. In this case, the threshold processing may be processing that uniformly changes pixels that have a predetermined first luminance value or less, the first luminance value being darker than the luminance values ​​of the pixels representing the plurality of vehicles, to a second luminance value that is darker than the first luminance value.

[0022] This allows background pixels that are not vehicles to have lower brightness values, and pixels that represent vehicles in the segmented image to be more emphasized.

[0023] (5) In the information processing device described in (1) to (4) above, the control unit may obtain the number of segments, which is the number of small regions included in the divided image, and may generate the divided image by dividing the SAR image according to the number of segments in accordance with the SLIC algorithm.

[0024] This allows a segmented image to be generated based on the SAR image.

[0025] (6) In the information processing device of (5) described above, the control unit may extract one independent vehicle that is not included in the set from the SAR image, and may obtain the number of segments based on a value obtained by dividing the number of pixels representing the independent vehicle from the total number of pixels in the SAR image.

[0026] This makes it possible to more reliably generate a divided image in which the vehicle is divided into separate small regions.

[0027] (7) In the information processing device according to any one of (1) to (4) above, the control unit may generate the divided images according to a watershed algorithm or a quickshift algorithm.

[0028] This allows a segmented image to be generated based on the SAR image.

[0029] (8) In the information processing device according to any one of (1) to (7) above, the control unit may output the number of the plurality of vehicles included in the set based on the divided image.

[0030] By counting based on the segmented images, the number of vehicles included in the collection can be determined more accurately than when the SAR image is used as is.

[0031] (9) The method according to this embodiment is an information generation method performed by the information processing devices described above in (1) to (8). Therefore, the information generation method according to this embodiment has the same effects as the information processing devices described above in (1) to (8).

[0032] (10) The computer program according to this embodiment is a computer program for causing a computer to function as the information generating device described above in (1) to (8). Therefore, the computer program according to this embodiment has the same effects as the information processing device described above in (1) to (8).

[0033] <Details of the Embodiments of the Present Invention> Hereinafter, the details of the embodiments of the present invention will be described with reference to the drawings. Note that at least some of the embodiments described below may be combined in any manner.

[0034] [Definition of Terms] Before describing the details of this embodiment, we will first define the terms used in this specification. "Vehicle": refers to all vehicles that travel on roads. The drive system of the vehicle is not limited to internal combustion engines, and electric vehicles and hybrid cars are also included in the vehicle. In this embodiment, when simply referring to a "vehicle," it includes both a probe vehicle that has an on-board device that can transmit probe information and a normal vehicle that does not have such an on-board device.

[0035] "Probe information" refers to information including various data about a vehicle sensed by the vehicle while it is traveling on a road. Probe information is also called probe data or floating car data. Probe information may include various vehicle attribute data such as vehicle identification information, vehicle position, vehicle speed, vehicle direction, and the time of occurrence of these data.

[0036] "Probe vehicle": A vehicle that can transmit probe information to an external device such as a server. More specifically, a vehicle that senses vehicle attribute data and transmits probe information including the sensed data to an external device.

[0037] "Synthetic aperture radar image": Data observed by a Synthetic Aperture Radar (SAR), in which pixel values ​​represent gray levels corresponding to the backscattering intensity at each point on the Earth's surface. Also called a "SAR image."

[0038] The SAR image may be any of a "single-polarized SAR image" that visualizes the backscattering intensity of a single polarized wave at a single point in time, a "dual-polarized SAR image" that visualizes the difference in scattering characteristics of each polarized wave at a single point in time, and an "intensity difference SAR image" that visualizes the change in the backscattering intensity of a single polarized wave. The product processing level of the SAR image may be any of "Single Look Complex (SLC) data," "Geocoded data," and "Ortho-processed data," but from the viewpoint of saving the user the trouble of correction processing, etc., higher level data is preferable.

[0039] "Optical image": Data in which the level of visible light at each point on the Earth's surface is used as a pixel value, as observed by optical satellites, aircraft, etc. In image processing at the information providing server 2, the data formats of optical images and SAR images are unified before processing is performed.

[0040] [System Configuration Example] Fig. 1 is a schematic diagram showing a configuration example of an information providing system 1. Fig. 2 is a block diagram showing a configuration example of an information providing server 2 and an in-vehicle device 4 included in the information providing system 1. Fig. 3 is a block diagram showing an internal configuration example of the information providing server 2.

[0041] 1 and 2, the information providing system 1 includes an information providing server 2, an on-board device 4 of a probe vehicle 3, a user terminal 6 carried by a user 5, an SAR image server 10, and an optical image server 11. Hereinafter, the information providing server 2 may be abbreviated as "server 2," and the probe vehicle 3 may be abbreviated as "vehicle 3." Furthermore, vehicles in general, regardless of whether they are equipped with an on-board device 4, will also be referred to as "vehicle 3" as appropriate.

[0042] The information providing system 1 of this embodiment is a system that distributes road traffic-related information (events) to an in-vehicle device 4 of a vehicle 3, a user terminal 6, etc. In the information providing system 1, an information providing server 2 is a type of information processing device that generates the information to be provided.

[0043] The provided information is information generated based on traffic conditions (described below) detected by the server 2. The provided information includes, for example, congestion information on roads, information on shoulder parking (hereinafter also referred to as "shoulder parking"), and parking lot status information. The provided information is generated by the server 2 using probe information, SAR images, and the like as source data.

[0044] The information providing server 2 is operated by an automobile manufacturer or an IT (Information Technology) company that operates various information distribution businesses. The information providing server 2 may be either an on-premise server or a cloud server. In response to a search request from the in-vehicle device 4 or the user terminal 6, the information providing server 2 can also execute a route search process based on a predetermined algorithm such as the Dijkstra algorithm or the potential method. In this case, the server 2 transmits the searched driving route to the sender of the search request.

[0045] The SAR image server 10 is operated by a private company or government agency that distributes SAR images of a specific area. The SAR images are generated by performing predetermined image processing on data received from a satellite equipped with a SAR. The SAR image server 10 may be either an on-premise server or a cloud server.

[0046] The SAR image may be an image acquired by observing the ground using an aircraft equipped with a SAR, and the aircraft is not limited to a satellite. For example, the aircraft may be a manned aircraft or an unmanned aircraft such as a drone.

[0047] The optical image server 11 is operated by a private company or government agency that distributes optical images of a specific area. Optical images are generated by performing predetermined image processing on data received from an optical satellite or aircraft. The optical image server 11 may be either an on-premise server or a cloud server. If a single organization distributes both SAR and optical images, the SAR image server 10 and the optical image server 11 may be the same server.

[0048] The in-vehicle device 4 of the vehicle 3 communicates wirelessly with wireless base stations 7 (e.g., mobile base stations) located in various locations. The wireless base stations 7 communicate with the information providing server 2 via a public communication network 8, which includes the Internet. Specifically, the in-vehicle device 4 wirelessly transmits uplink information S1 addressed to the server 2 to the wireless base station 7. The uplink information S1 reaches the server 2 from the wireless base station 7 via the public communication network 8. The server 2 also transmits downlink information S2 addressed to a specific in-vehicle device 4 to the public communication network 8. The downlink information S2 reaches the specific in-vehicle device 4 from the public communication network 8 via the wireless base station 7.

[0049] The user terminal 6 is a data communication terminal carried by the user 5, such as a smartphone, tablet computer, or laptop computer. The user terminal 6 communicates wirelessly with wireless base stations 7 in various locations. Therefore, the user terminal 6 wirelessly transmits uplink information S1 addressed to the server 2 to the wireless base station 7. The information providing server 2 transmits downlink information S2 addressed to the specific user terminal 6 to the public communication network 8.

[0050] 1 and 2 illustrate a mobile terminal as the user terminal 6. However, the user terminal 6 may also be a stationary terminal device such as a desktop computer installed indoors. In this case, the user terminal 6 may communicate with the information providing server 2 via a fixed communication network that is connected to the public communication network 8, such as an optical communication line. The information providing server 2 may also transmit various types of information to another server computer (not shown) that communicates via the public communication network 8, etc.

[0051] [Configuration of Information Providing Server] See Fig. 3. The information providing server 2 is a server computer including a control unit 21, a storage unit 22, and a communication unit 23. The storage unit 22 includes a storage area in which a plurality of databases 24, 25, 26, 27, 28, and 29 are constructed.

[0052] Here, the information providing server 2 may be configured with one computer or multiple computers. When configured with multiple computers, these multiple computers may be installed in the same facility or may be scattered across multiple geographically separated locations. When the information providing server 2 is configured with multiple computers, the functions of the information providing server 2 may be realized by the multiple geographically separated computers cooperating with each other via a network such as the public communication network 8.

[0053] The control unit 21 is an arithmetic processing device including a CPU (Central Processing Unit) and a RAM (Random Access Memory). The control unit 21 may include an integrated circuit other than the CPU, such as an FPGA (Field-Programmable Gate Array). The control unit 21 reads a computer program P1 stored in the storage unit 22 into memory (RAM) and performs predetermined information processing in accordance with the read computer program P1.

[0054] The storage unit 22 is an auxiliary storage device including a nonvolatile memory such as a hard disk drive (HDD) and a solid state drive (SSD). The storage unit 22 may also include a flash read-only memory (ROM), a universal serial bus (USB) memory, or an SD card.

[0055] The computer program P1 includes a program for calculating link travel times and congestion lengths, a route search program, and other programs for causing the CPU to execute the processes of generating the segmented images shown in FIG. 4, detecting traffic conditions, and various distribution processes.

[0056] The communication unit 23 is a communication interface that communicates with the wireless base station 7 via the public communication network 8. The communication unit 23 can receive uplink information S1 that the wireless base station 7 has transmitted to the device itself, and can transmit downlink information S2 that the device itself has generated to the wireless base station 7.

[0057] The communication unit 23 is also connected to the SAR image server 10 and the optical image server 11 via the public communication network 8. The communication unit 23 periodically receives SAR image data and optical image data from the SAR image server 10 and the optical image server 11. The communication unit 23 transfers the received SAR image data and optical image data to the control unit 21.

[0058] The multiple databases 24, 25, 26, 27, 28, and 29 are constructed in large-capacity storage such as an HDD or SSD included in the storage unit 22. The large-capacity storage of the storage unit 22 may be one or more external storage devices connected to the information providing server 2. At least one of the multiple databases 24, 25, 26, 27, 28, and 29 may be constructed in the storage of a cloud server operated by an IT company other than the information providing server 2.

[0059] In this way, the memory unit 22 in the sense of storage in which multiple databases 24, 25, 26, 27, 28, and 29 are constructed may be, from the perspective of the information providing server 2, a memory unit of its own device, or an external device (such as the above-mentioned external device or cloud server).

[0060] The plurality of databases 24 , 25 , 26 , 27 , 28 , 29 include a map database 24 , an optical image database 25 , a coverage database 26 , a SAR image database 27 , a segmented image database 28 , and a detected object database 29 .

[0061] The map database 24 includes, for example, digital road map data (hereinafter referred to as "DRM") covering the entire country. The DRM includes "intersection data" and "link data." The "intersection data" is data that associates intersection IDs assigned to domestic intersections with location information of the intersections. The "link data" is data that associates the following information a to e with link IDs of specific links assigned to domestic roads.

[0062] Information a: Position information of the start point, end point, and interpolation point of a specific link. Information b: Link ID connected to the start point of a specific link. Information c: Link ID connected to the end point of a specific link. Information d: Link cost of a specific link.

[0063] The DRM forms a network corresponding to the actual road alignment and driving direction of the road. In other words, the DRM is a network in which road sections between nodes representing intersections are connected by directed links l (lowercase L). More specifically, the DRM is a directed graph in which a node n is set for each intersection and each node n is connected by a pair of directed links l in opposite directions. Therefore, in the case of a one-way road, only one-way directed links l connect the node n.

[0064] The DRM also includes road attribute information of the road corresponding to the directed link l. The road attribute information includes, for example, the following information 1 to 6: Information 1: Road type information indicating whether the road is an ordinary road or a toll road Information 2: Average gradient information of the road Information 3: Number of lanes on the road Information 4: Road width information for each lane 5: Radius of curvature information of the road Information 6: Regulated speed limit of the road (for example, legal speed limit)

[0065] The optical image database 25 stores optical images of a predetermined area received at a predetermined frequency from the optical image server 11. When an optical image of the same area is acquired, it is successively updated with the latest optical image.

[0066] The target range database 26 stores a detection target range A1. The detection target range A1 is a geographical area in which traffic condition detection processing is performed based on SAR images. Specifically, the detection target range A1 is a range that includes polygon data PD1 and PD2 that represent the outer limits of areas within which the presence of a vehicle 3 can be determined from the air, such as roads and parking lots. In the example of FIG. 3 , PD1 represents polygon data for roads, and PD2 represents polygon data for parking lots.

[0067] The SAR image database 27 stores SAR images SIi (i = 1, 2 ...) received at a predetermined frequency from the SAR image server 11. The subscript i of the symbol SI representing the SAR image is an identification value that indicates the oldest SAR image created by the SAR image server 11, with a larger value indicating a newer image created. Furthermore, if the server 2 acquires multiple SAR images SI from the same region but created at different times, the multiple SAR images created at different times are archived as time-series data.

[0068] The divided image database 28 stores divided images DIi (i = 1, 2 ...). The divided images DIi are generated based on the SAR images SIi. More specifically, the control unit 21 generates the divided images DIi by grouping the SAR images SIi corresponding to the detection target range A1 into small, irregularly shaped regions in a divided image generation process (step ST13) described below. The subscript i in the symbol DI representing a divided image corresponds to the subscript i in the symbol SI representing a SAR image. For example, a divided image generated based on a predetermined SAR image SI1 will be described using the symbol DI1.

[0069] Information about detected objects detected based on divided images or difference images in the detection target range A1 is stored in the detected object database 29. A difference image is, for example, an image that shows the difference between a plurality of divided images.

[0070] The information about the detected object includes, for example, the type of the detected object, time information, and shape information. The type of the detected object includes, for example, a vehicle, a line of vehicles, a parked vehicle on the side of the road, and a missing part of the road. The time information is, for example, the time when a state change occurs in the detected object. The time information may be the time when the detected object is detected. The shape information of the detected object includes, for example, the following information depending on the type of the detected object:

[0071] Information M1: Coordinates (e.g., latitude and longitude) of the center position of an object considered to be a vehicle 3. Information M2: Line data connecting the start and end points of a group of objects considered to be a convoy, and the number of vehicles 3 included in the convoy. The line data may be, for example, coordinates (e.g., latitude and longitude) of the start and end positions, or the extension direction and distance from the start and end positions. Information M3: Line data connecting the start and end points of an event considered to be roadside parking. Information M4: Polygon data representing the range considered to be missing from the road.

[0072] A member database (not shown) is also created in the memory unit 22. The member database includes personal information such as the addresses and names of registered members. Registered members include, for example, owners of vehicles 3, owners of user terminals 6, and operation managers of other servers. The member database may further include identification information (e.g., at least one of MAC addresses, email addresses, and telephone numbers) of the communication devices (e.g., the in-vehicle device 4 and the user terminal 6) of registered members.

[0073] 2, the in-vehicle device 4 is an in-vehicle network including an ECU (Electric Control Unit). The in-vehicle device 4 includes a processing unit 31, a storage unit 32, and a communication unit 33.

[0074] The storage unit 32 is a storage device that includes at least one nonvolatile memory (storage medium) of an HDD and an SSD, and a volatile memory (storage medium) such as a random access memory. The nonvolatile memory of the storage unit 32 stores a computer program 34. The computer program 34 includes a driving control program to be executed by the processing unit 31, a route search program for the probe vehicle 3, and an image processing program for displaying the searched route on the display of the navigation device.

[0075] The processing unit 31 is a processing unit including a CPU. The processing unit 31 reads out a computer program 34 stored in the nonvolatile memory of the storage unit 32 and performs various information processing in accordance with the computer program 34.

[0076] The communication unit 33 is a wireless communication device permanently mounted on the probe vehicle 3 or a data communication terminal (for example, a smartphone, a tablet computer, or a node-type personal computer) temporarily mounted on the probe vehicle 3. The communication unit 33 has a GNSS (Global Navigation Satellite System) receiver. The processing unit 31 monitors the current position of the vehicle in almost real time based on the GNSS position information received by the communication unit 33.

[0077] The processing unit 31 records various vehicle attribute data, such as vehicle speed, vehicle direction, steering angle, engine RPM, brake pressure, and accelerator pedal position, measured at a predetermined sensing period, together with the vehicle position and the measurement time, in the storage unit 32. When the vehicle attribute data for a predetermined period (e.g., 10 seconds) has been accumulated in the storage unit 32, the communication unit 33 generates probe information including the accumulated vehicle attribute data and identification information of the vehicle itself, and transmits the generated probe information via uplink to the information providing server 2.

[0078] The on-board device 4 of the probe vehicle 3 further includes an input interface that accepts operational inputs from the driver and a display that presents map images corresponding to the digital road map data (DRM) to the driver. The input interface is, for example, an input device associated with the navigation device. The input interface may also be an input device of a data communication terminal mounted on the probe vehicle 3. The display is, for example, a liquid crystal display associated with the navigation device. The display may also be a head-up display that projects an image onto the glass surface in front of the driver's seat.

[0079] 4 is a flowchart illustrating a series of processes executed by the control unit 21 of the information providing server 2. The control unit 21 executes the following various processes based on a computer program P1 stored in the storage unit 22.

[0080] The control unit 21 executes the following processes. Note that, among the following processes, it is possible to execute only one of the traffic condition distribution process (step ST15) and the forecast information distribution process (step ST16).

[0081] Step ST11: Extraction processing of vehicle existence range Step ST12: Generation processing of detection target range Step ST13: Generation processing of divided image Step ST14: Detection processing of traffic condition Step ST15: Distribution processing of traffic condition Step ST16: Distribution processing of prediction information

[0082] (Vehicle Existence Area Extraction Process) First, the control unit 21 extracts a vehicle existence area as an area where the vehicle 3 may be present from the digital road map data (DRM) of a predetermined area read from the map database 24 (step ST11).

[0083] The extraction process in step ST11 is a process of generating polygon data representing the outline of a road based on, for example, the position information of a specific link in the DRM, the number of lanes, and the road width of each lane. Note that if the range of a parking lot is included in the background data of the DRM, polygon data of the parking lot may be generated.

[0084] (Detection target range generation process) Next, the control unit 21 generates a detection target range A1 (step ST12) based on the vehicle presence range extracted in step ST11 and the optical image data of the predetermined area read from the optical image database 25. The control unit 21 stores the generated detection target range A1 in the target range database 26.

[0085] The generation process of step ST12 includes, for example, the following processes: Process 1: Read out optical image data of a specified area. Process 2: Identify, from the optical image data, objects to be excluded from the detection target range A1. Examples of objects to be excluded include medians, shrubs, traffic lights, signs, utility poles, the shadows of buildings, and pedestrian bridges. The control unit 21 identifies these objects to be excluded by inputting the optical image data into a trained learning device that is capable of recognizing these objects to be excluded. Process 3: Delete the objects to be excluded from the optical image data, and generate polygon data representing the outline of the road and polygon data representing the outline of the parking lot based on the optical image data.

[0086] Process 4: If there is a discrepancy in shape or position between the polygon data based on the DRM and the polygon data based on the optical image data, correct the shape and position to match the latter. Process 5: Store the corrected polygon data in the target range database 26 as polygon data PD1 representing the outline of the road and polygon data PD2 representing the outline of the parking lot.

[0087] (Segmented Image Generation Process) Next, the control unit 21 generates segmented images based on the SAR image (step ST13). Specifically, the control unit 21 generates segmented images only for areas of the SAR image read from the SAR image database 27 that correspond to the polygon data PD1 and PD2 read from the target range database 26. This configuration makes it possible to limit the areas for which segmented images are generated, and by reducing the amount of information processing in the control unit 21, it is possible to shorten the time required to generate segmented images. Note that the control unit 21 may also generate segmented images for the entire SAR image.

[0088] Fig. 5 is a flowchart showing detailed steps of the segmented image generation process. Fig. 6 is an example of an SAR image used in the segmented image generation process. Fig. 7 is a schematic diagram showing how segmented images are generated from an SAR image. Below, the process of generating segmented image DI1 based on SAR image SI1 shown in Fig. 6 will be described with appropriate reference to Figs. 5 to 7.

[0089] As shown in FIG. 6 , the SAR image SI1 includes a region R1 corresponding to the polygon data PD1 and a region R2 corresponding to the polygon data PD2. The control unit 21 generates segmented images only for these regions R1 and R2. The SAR image SI1 also includes sets L1 and L2 of multiple vehicles 3 positioned close to each other. These sets L1 and L2 are, for example, convoys in which multiple vehicles 3 are lined up in one or more rows, and will be referred to below as "vehicle convoys L1" and "vehicle convoys L2." Note that the sets of vehicles 3 for which the control unit 21 generates segmented images are not limited to sets lined up on a roadway or in a parking lot, such as the convoys L1 and L2 described below, but may also be sets of vehicles 3 arranged irregularly as long as they are spatially close to each other.

[0090] Vehicle train L1 is a line of multiple vehicles 3 included in region R1, such as a group of vehicles waiting at a traffic light on the road. Vehicle train L2 is a line of multiple vehicles 3 included in region R2, such as a group of vehicles parked in a parking lot. Vehicle trains L1 and L2 may be multiple vehicles 3 lined up in a single line, as shown in vehicle train L1, or multiple vehicles 3 lined up in two or more lines, as shown in vehicle train L2.

[0091] Furthermore, the SAR image SI1 includes one independent vehicle 3 that is not included in either of the vehicle trains L1 and L2. Such an independent vehicle 3 will be referred to as an "independent vehicle X1" as appropriate. The independent vehicle X1 is, for example, a vehicle 3 traveling on a road in the region R1, and is separated from the other vehicles 3 by a distance equal to or greater than a predetermined inter-vehicle distance D1. The inter-vehicle distance D1 is, for example, the distance of one vehicle, e.g., 5 m. Conversely, the vehicles 3 included in the vehicle trains L1 and L2 are adjacent to the other vehicles 3 at a distance narrower than the inter-vehicle distance D1.

[0092] For ease of explanation, SAR image SI1 shows vehicle 3 and vehicle lines L1 and L2, but at the stage when SAR image SI1 is acquired, control unit 21 is unable to recognize whether vehicle 3 and the like included in SAR image SI1 are "vehicles" or "other objects" installed on the road.

[0093] 5, the control unit 21 executes the following processes as the process of generating the divided image: Step ST21: Threshold value processing Step ST22: Processing for obtaining the number of segments Step ST23: Division processing

[0094] First, the control unit 21 performs threshold processing on the SAR image SI1 read out from the SAR image database 27 (step ST21). Specifically, the control unit 21 performs processing to uniformly change pixels in the SAR image SI1 that have a predetermined first luminance value B1 or less, which is darker than the luminance values ​​of the pixels representing the plurality of vehicles 3, to a second luminance value B2 that is darker than the first luminance value B1.

[0095] As a result, as shown in (a) and (b) in Figure 7, pixels in the SAR image SI1 that have a first luminance value B1 or less are replaced with a second luminance value B2, which increases the difference in luminance value between the vehicle 3 and the background in the SAR image SI1 after threshold processing, thereby emphasizing the vehicle 3 and improving the accuracy of the segmentation process described below.

[0096] In regions R1 and R2, pixels that show the vehicle 3 are rendered brighter than pixels that do not show the vehicle 3 (for example, pixels that show an asphalt road). For this reason, for example, if the first luminance value B1 is set to "75," which is half the average luminance value of the pixels that represent the vehicle 3 (for example, a luminance value of about 150), and the second luminance value B2 is set to "0," the control unit 21 can maintain the luminance values ​​of the pixels that represent the vehicle 3 in regions R1 and R2, while uniformly changing the luminance values ​​of pixels that are equal to or less than the first luminance value B1 (= 75) to "0." This allows background pixels that are not the vehicle 3 to have a lower luminance value, thereby emphasizing the pixels that represent the vehicle 3.

[0097] The first luminance value B1 and the second luminance value B2 are stored in the storage unit 22 as preset values. The first luminance value B1 and the second luminance value B2 may be values ​​set by an operator based on empirical rules, or may be values ​​calculated by the control unit 21 based on the average of the luminance values ​​in regions R1 and R2. In this case, the first luminance value B1 and the second luminance value B2 may be values ​​prepared for each of regions R1 and R2. For example, a value calculated based on the average of the luminance values ​​in region R1 may be set as the first luminance value B11 used in the threshold processing of region R1, and a value calculated based on the average of the luminance values ​​in region R2 may be set as the first luminance value B12 used in the threshold processing of region R2.

[0098] Next, the control unit 21 acquires the number of segments (step S22). The number of segments is the number of small regions into which the regions R1 and R2 are divided in the division process described below. The greater the number of segments, the smaller the regions R1 and R2 are divided into.

[0099] There are two methods for the control unit 21 to obtain the number of segments. The first method is to read out the number of segments stored in the storage unit 22 as a preset value from the storage unit 22. In this case, the number of segments may be a default value stored in advance in the storage unit 22, or may be a value obtained in accordance with the SAR image SIi by calculating the total number of pixels and spatial resolution obtained in accordance with the SAR image SIi with a coefficient stored in advance in the storage unit 22.

[0100] For example, if the total number of pixels included in region R1 is 18,000 and the spatial resolution of SAR image SI1 is 0.5 m, the actual area of ​​the ground surface represented by region R1 is 0.25 square meters per pixel, and the total area of ​​region R1 is 4,500 square meters (= 18,000 × 0.25). If a typical area of ​​vehicle 3 is 10 square meters (total length 5 m, total width 2 m), the minimum number of segments (minimum number of segments) required to fill in the pixels corresponding to vehicle 3, which is the object to be detected, is 450 (= 4,500 / 10).

[0101] The control unit 21 performs the above calculation. For example, a representative area value of the vehicle 3 (e.g., 10) is stored in advance in the storage unit 22 as a coefficient. Next, the control unit 21 acquires the total number of pixels in the region R1 and the spatial resolution according to the SAR image SI1. The control unit 21 then acquires the number of segments by dividing the product of the total number of pixels in the region R1 and the square of the spatial resolution by the representative area value of the vehicle 3. At this time, the acquired number of segments may be the minimum number of segments (450 in the above example) or a value greater than the minimum number of segments.

[0102] For example, the number of segments may be the minimum number of segments plus a predetermined value (e.g., 450 + α). Alternatively, the number of segments may be the minimum number of segments multiplied by a predetermined value greater than 1 (e.g., 450 × β). By obtaining a number equal to or greater than the minimum number of segments as the number of segments, it is possible to more reliably generate a divided image in which the vehicle 3 is divided into separate small regions in the division process described below.

[0103] A second method for the control unit 21 to obtain the number of segments is a method in which the control unit 21 automatically calculates the number of segments based on the SAR image SI 1. For example, the control unit 21 extracts one independent vehicle X1 that is not included in the vehicle continuities L1 and L2 from the SAR image SI 1.

[0104] As an extraction method, for example, a specific region XR in the SAR image SI1 where the independent vehicle X1 is likely to appear may be set in advance, and a pixel region brighter than a predetermined third luminance value B3 in the specific region XR may be extracted as the independent vehicle X1. The specific region XR may be, for example, a region of a road downstream of a traffic light-activated intersection that is close to the intersection. In such a specific region XR, the vehicle 3 that has passed through the intersection is likely to be present alone, allowing for more accurate extraction of the independent vehicle X1.

[0105] The control unit 21 then obtains the number of segments based on the value obtained by dividing the number of pixels representing the independent vehicle X1 from the total number of pixels in the SAR image SI1 (if the segmentation process is performed for each of the regions R1 and R2, the total number of pixels for each of the regions R1 and R2). For example, the control unit 21 obtains the number of segments by dividing the number of pixels representing the independent vehicle X1 (e.g., 40 pixels) from the total number of pixels in the region R1 of the SAR image SI1 (18,000 pixels in the above example). By calculating the number of segments based on the SAR image SI1 in this way, it is possible to obtain a more appropriate number of segments depending on various SAR images.

[0106] Next, the control unit 21 executes a segmentation process (step ST23). For example, the control unit 21 generates a segmented image DI1 in which the pixels representing the plurality of vehicles 3 are segmented into separate small regions by grouping a plurality of pixels in the SAR image SI1 that have similar distances and luminance values ​​into small regions of an irregular shape.

[0107] For example, when the minimum number of segments is used as the number of segments, the divided image DI1 generated by the control unit 21 may be an image in which each of the multiple vehicles 3 is divided into one small region. Furthermore, when twice the minimum number of segments is used as the number of segments, the divided image DI1 may be an image in which each of the multiple vehicles 3 is divided into two small regions.

[0108] By dividing multiple vehicles 3 into separate small regions, it is possible to distinguish between vehicles 3 included in a vehicle line and other objects in the SAR image SIi, thereby enabling more appropriate vehicle recognition.

[0109] More specifically, the control unit 21 generates a segmented image DI1 by dividing the SAR image SI1 into multiple small regions according to various parameters using the Simple Linear Iterative Clustering (SLIC) algorithm. Small, irregularly shaped regions in which multiple pixels are grouped together through the segmentation process are also called "superpixels." The SLIC algorithm is a method that incorporates k-means to generate superpixels.

[0110] Here, the various parameters used in the SLIC algorithm include the number of segments (also referred to as the "k value") acquired in step ST22, the r value, and the m value. The r value and the m value are stored in advance in the memory unit 22, for example. The r value is a smoothing coefficient for smoothing the luminance values ​​of pixels included in the same superpixel. The m value is a value that balances the similarity of luminance values ​​and the proximity of pixels. The smaller the m value, the more the similarity of luminance values ​​is emphasized, and the larger the m value, the more the proximity of pixels is emphasized. Note that the m value may be automatically determined by another algorithm, such as the SLICO algorithm.

[0111] 7(b) and 7(c), a divided image DI1 is generated in which a plurality of pixels having similar luminance values ​​are grouped into superpixels. The control unit 21 stores the generated divided image DI1 in the divided image database 28.

[0112] As shown in (c) of Figure 7, the divided image DIi includes multiple small regions SP1 and SP2. The pixels included in the multiple small regions SP1 and SP2 are smoothed based on the r value. Because the luminance value of the pixels showing the vehicle 3 is brighter than the luminance value of the pixels in other regions showing the road, etc., the small region SP1 including the vehicle 3 is highlighted brighter than the small region SP2 not including the vehicle 3. This allows the vehicle 3 to be recognized more appropriately.

[0113] The control unit 21 may generate the segmented image according to a superpixel generation algorithm other than the SLIC algorithm. For example, the control unit 21 may generate the segmented image according to the watershed algorithm or the quickshift algorithm. Any of these methods can be used to generate a segmented image appropriately based on the SAR image. These superpixel generation algorithms are sometimes collectively referred to as "segmentation methods."

[0114] Here, if a pixel representing one vehicle 3 in the SAR image SI1 contains a dark area, the pixel representing one vehicle 3 may be divided into multiple particles by threshold processing (step ST21), and the one vehicle 3 may be divided into separate small areas in the subsequent division processing (step ST23).

[0115] To prevent this, the control unit 21 may execute the following process. For example, parameters related to the standard size of the vehicle 3 (hereinafter referred to as "standard size") are stored in advance in the storage unit 22. Then, during the division process, the control unit 21 executes a process of combining multiple small regions that are close in distance (e.g., adjacent to each other) and each have a size less than the standard size so that the size approximates the standard size (e.g., the number of pixels is in the range of 70% to 130% of the standard size). In this way, even if one vehicle 3 is divided into multiple small regions due to threshold processing, these can be combined to form one vehicle 3 into a single small region, thereby enabling more accurate recognition of the vehicle 3.

[0116] (Traffic Condition Detection Process) See Fig. 4. Next, the control unit 21 detects a traffic condition based on the segmented image DIi read from the segmented image database 28 (step ST14). The traffic condition is information about road traffic acquired based on changes such as the appearance, increase, or decrease of detected objects (e.g., vehicles, convoys, shoulder parking, and missing roads).

[0117] Fig. 8 is a flowchart showing an example of a traffic condition detection process. Fig. 9 is a flowchart showing another example of a traffic condition detection process. As the traffic condition detection process, the control unit 21 executes a process of detecting a traffic condition based on one divided image DIi (Fig. 8) and a process of detecting a traffic condition based on the difference between a plurality of divided images DIi (Fig. 9).

[0118] See FIG. 8. The control unit 21 extracts the contour of the vehicle 3 from one divided image DIi (contour extraction process: step ST31). Specifically, the control unit 21 applies a dilation filter and a contraction filter to the divided image DIi as preprocessing to fill in small noise areas contained in the target region, and then extracts the contour using an appropriate edge detection process. The specific method of edge detection is not particularly limited, but the contour may be extracted using, for example, the Canny algorithm. Alternatively, the contour may be extracted by performing binarization using a predetermined threshold value. As a result, a contour image in which the contour of the vehicle 3 is emphasized is generated, as shown in (c) and (d) in FIG. 7.

[0119] Next, the control unit 21 detects an object such as a vehicle 3 based on the divided image DIi (object detection process: step ST32). Specifically, the control unit 21 inputs small areas included in the divided image DIi into a trained learning device that can classify them into "vehicle 3," "road defect," "signboard," etc., and recognizes the small areas that have been assigned these classifications as pixels displaying a detected object. The control unit 21 records information about the detected object in the detected object database 29.

[0120] The control unit 21 may detect the "vehicle 3" by more detailed classification, such as a large truck (e.g., a total length of approximately 12 m), a standard car (e.g., a total length of 4 to 5 m), or a light car (e.g., a total length of approximately 3.4 m). In this case, the control unit 21 may detect the object based on information regarding the number of small areas in addition to the divided image DIi itself, such as classifying a vehicle 3 appearing across two small areas SP1 as a large truck and a vehicle 3 appearing in one small area SP1 as a standard car or a light car.

[0121] Finally, the control unit 21 counts the area enclosed by these contours and outputs the result (counting process: step ST33). The specific counting method is not particularly limited, but the number of contours may be output, for example, by using the LEN function in Python. Alternatively, the number of contours may be output by using a labeling process such as connectedComponents in OpenCV. In the case of (d) in FIG. 7 , the control unit 21 outputs "12" as the number of vehicles 3 included in the convoy. The control unit 21 records the output number in the detected object database 29 as one piece of information indicating the detected object (specifically, information M2).

[0122] By counting based on the segmented images, the number of vehicles 3 included in the vehicle train can be determined more accurately than when the SAR image is used as is.

[0123] 9 , the control unit 21 acquires a difference image based on a plurality of divided images DIi (difference processing: step ST41). For example, the control unit 21 acquires a difference image by calculating the difference between a divided image DIm and a divided image DIn (where m<n) that is newer than the divided image DIm.

[0124] Here, of the two segmented images DIm, DIn, the newer segmented image DIn is, for example, the newest segmented image DIi stored in the segmented image database 28. In other words, it is the segmented image DIk generated based on the SAR image SIk acquired at a generation time point k (hereinafter referred to as the "most recent time point k") that has the smallest time difference from the time point at which the control unit 21 detects the traffic condition. By using such a newest segmented image DIk, the latest traffic condition can be detected.

[0125] However, depending on the progression cycle of the generation time point (for example, every day, every hour, every 10 minutes, etc.) and the type of event to be detected, the newer divided image DIn may be, for example, the divided image DI(k-1) from the generation time point k-1 just before the most recent time point k.

[0126] Of the two divided images DIm, SIn, the older divided image DIm is, for example, the divided image DI(k-1) that is one image before the most recent time point k. However, depending on the progression cycle of the generation time point and the type of event to be detected, for example, the older divided image DIm may be the divided image DIm that is two or more images before the most recent time point k (m≦k-2).

[0127] Next, the control unit 21 detects traffic conditions based on the differential image (condition detection process: step ST42). Examples of traffic conditions, i.e., changes in detected objects (hereinafter referred to as "condition changes") include "appearance of a vehicle 3," "changes in the line of vehicles," "parking on the shoulder of the road," and "missing road."

[0128] For example, if the detected state change is “the appearance of a vehicle 3 ,” the control unit 21 records the coordinates of the center position of the vehicle 3 and the time information of the state change in the detected object database 29 .

[0129] Furthermore, if the detected state change is a "change in the vehicle convoy," the control unit 21 records the line data connecting the start point and end point of the vehicle convoy and the change in the number of vehicles 3 included in the vehicle convoy in the detected object database 29.

[0130] Here, a change in a vehicle stream refers to either the occurrence of a vehicle stream, an increase in the number of vehicles, or a decrease in the number of vehicles. The occurrence of a vehicle stream refers to a state in which a vehicle stream is detected in the divided image DIn in an area of ​​the divided image DIm where no vehicle stream has been detected. In other words, the occurrence of a vehicle stream from a state in which there was no vehicle stream is referred to as "the occurrence of a vehicle stream."

[0131] Furthermore, "increase in the vehicle convoy" refers to a state in which a vehicle convoy has already occurred in the divided image DIm and the number of vehicles 3 included in the vehicle convoy has increased in the divided image DIn. For example, the control unit 21 calculates the amount of change in the number of vehicles included in the vehicle convoy based on the number of vehicles 3 output in the counting process (step ST33). If the number of vehicles 3 included in a given vehicle convoy in the divided image DIm is 12, and the number of vehicles 3 included in the vehicle convoy in the same area of ​​the divided image DIn is 20, the control unit 21 calculates the difference, "+8," as the amount of change. In this case, the vehicle convoy has increased from 12 to 20, so the control unit 21 detects "increase in the vehicle convoy" as the traffic state.

[0132] For example, when the control unit 21 detects an "increase in the line of vehicles" of a predetermined number or more in the region R1 (i.e., the region in which the vehicle 3 travels, e.g., a road), the control unit 21 may detect this state as "occurrence of traffic congestion." Furthermore, when the control unit 21 detects an "increase in the line of vehicles" of a predetermined number or more in the region R2 (e.g., a parking lot), the control unit 21 may detect this state as "the parking lot is full."

[0133] Conversely, if the number of vehicles 3 in the vehicle queue is decreasing from divided image DIm to divided image DIn, the control unit 21 detects a "decreasing number of vehicles in the vehicle queue" as the traffic state. For example, if the control unit 21 detects a "decreasing number of vehicles in the vehicle queue" in region R1 by a predetermined number or more, the control unit 21 may detect this state as a "clearance of traffic congestion." Furthermore, if the control unit 21 detects a "decreasing number of vehicles in the vehicle queue" in region R2 by a predetermined number or more, the control unit 21 may detect this state as a "vacant parking lot."

[0134] If the detected state change is "shoulder parking," the control unit 21 records line data connecting the start point and end point of the shoulder parking and time information of the state change in the detected object database 29. Furthermore, if the detected state change is "loss of road," the control unit 21 records polygon data representing the range of the loss and time information of the state change in the detected object database 29.

[0135] (Traffic Condition Distribution Processing) See Fig. 4. After the traffic condition detection processing (step ST14), the control unit 21 distributes the latest traffic conditions recorded in the detected object database 29 (traffic condition distribution processing: step ST15). Specifically, every time a new traffic condition is stored in the detected object database 29 by the traffic condition detection processing, the type, type information, and time information of the stored traffic condition are transmitted to the in-vehicle device 4 and the user terminal 6.

[0136] (Prediction Information Distribution Process) After the traffic condition detection process (step ST14), the control unit 21 creates statistical information on traffic conditions by time period for each predetermined road or parking lot, and distributes the created statistical information as prediction information (prediction information distribution process: step ST16). Specifically, when a predetermined type of traffic condition, such as a traffic congestion, is detected with a high frequency during a predetermined time period, prediction information including morphology information indicating the section where the traffic congestion occurs and the time period during which the traffic congestion is likely to occur is transmitted to the in-vehicle device 4 and the user terminal 6 as a possible future event.

[0137] At least one of the recommended destinations and routes may be provided to the in-vehicle device 4, the user terminal 6, etc. In this case, by generating the latest traffic conditions and forecast information acquired based on the divided image DIi, it is possible to suggest to the driver of the vehicle 3 parking lots that are predicted to be vacant and routes that are comfortable to travel to those parking lots.

[0138] [Example of Road Map Screen Display] Fig. 10 is an explanatory diagram showing an example of a road map screen display including events provided by the information providing server 2. The road map in Fig. 10 is a road map displayed on the display of the in-vehicle device 4 or the user terminal 6 that has received the provided information from the information providing server 2. As shown in Fig. 10, events related to road traffic provided by the server 2 include, for example, the following events:

[0139] Event E1: Roadside parking Event E2: Obstacle present Event E3: Possibility of road collapse Event E4: Expressway exit congestion Event E5: Parking lot full Event E6: Parking lot vacant Event E7: Congestion occurring section

[0140] Of the above events, the event E7 indicated by diagonal hatching is an event generated based on probe information, while the other events E1 to E6 are events generated based on divided images.

[0141] Here, since the probe information is data representing the travel trajectory of the probe vehicle 3 while it is traveling, it is difficult to determine from the probe information whether the vehicle is parking on the shoulder of the road (event E1), where the engine is stopped and the vehicle is parked, whether there are obstacles (event E2), or whether the parking lot is in good condition (events E5 and E6).

[0142] In contrast, the information providing server 2 of this embodiment can detect changes in the state of the divided images that occur in the detection range A1 (such as the appearance of a vehicle 3, the appearance of a traffic jam, the elimination of the traffic jam, a full parking lot, an empty parking lot, shoulder parking, and the absence of a road). Therefore, events E1 to E6 that cannot be generated when the original data is probe information can be generated from the divided images, and the generated events E1 to E6 can be provided to the user.

[0143] 10, the following events may be detected based on the difference between the divided images, and the detected events may be used as provided information: Event E8: A signboard (e.g., a construction sign) Event E9: An impassable section such as an icy section, a snow-covered section, or a flooded section Event E9 can be detected, for example, by determining whether the reflection intensity of the entire road has significantly decreased.

[0144] Event E10: stranded vehicle. Event E10 can be detected, for example, by determining whether a vehicle 3 is present on the road for a long period of time during bad weather such as snow or heavy rain. Event E11: recommended parking lot. Event E11 can be generated, for example, by determining the vacant parking lot with the fewest number of parked vehicles. Event E12: congestion at facility entrance / exit. Event E12 can be detected, for example, by determining whether the upstream position of the congestion is located near the entrance / exit of a specified facility such as a suburban store.

[0145] Event E13: A puddle on the road. Event E13 can be detected, for example, by determining whether or not an area with reduced reflection intensity exists over a predetermined range or more in a predetermined road section. Event E14: A flooded section of the road. Event E14 can be detected, for example, by determining whether or not a section with reduced reflection intensity exists over a predetermined length or more in a predetermined road section.

[0146] [Other Modifications] The embodiments disclosed herein are illustrative in all respects and are not restrictive. The scope of the present invention is not limited to the above-described embodiments, but includes all modifications within the scope of the claims and equivalents thereof.

[0147] In the above-described embodiment, the control unit 21 may create the detection target range A1 from only the digital road map (DRM) or only the optical image. Alternatively, the optical image server 11 may generate the detection target range A1. In this case, the information providing server 2 may record the detection target range A1 acquired from the optical image server 11 in its own target range database 26.

[0148] REFERENCE SIGNS LIST 1 Information provision system 2 Information provision server (server) 3 Probe vehicle (vehicle) 4 In-vehicle device 5 User 6 User terminal 7 Wireless base station 8 Public communication network 10 Image server 11 Optical image server 21 Control unit 22 Memory unit (database storage) 23 Communication unit 24 Map database 25 Optical image database 26 Target range database 27 Image database 28 Segmented image database 29 Detected object database 31 Processing unit 32 Memory unit 33 Communication unit 34 Computer program P1 Computer program S1 Uplink information S2 Downlink information DRM Digital map data A1 Detection target range PD1 Polygon data (road) PD2 Polygon data (parking lot) SIi SAR image DIi Segmented image R1 Area R2 Area XR Specific area L1 Vehicle convoy L2 Vehicle convoy X1 Independent vehicle D1 Inter-vehicle distance B1 First luminance value B11 First luminance value B12 First luminance value B2 Second luminance value B3 Third luminance value SP1 Small area SP2 Small area

Claims

1. An information processing device comprising: a control unit that generates a segmented image in which a group of vehicles located close to each other are divided into separate small regions by grouping similar pixels from an SAR image containing the group of vehicles located close to each other into small, irregularly shaped regions.

2. The information processing device according to claim 1, wherein the control unit groups a plurality of pixels in the SAR image that are close in distance into the small region.

3. The information processing device according to claim 2, wherein the control unit groups a plurality of pixels in the SAR image that have similar distances and luminance values ​​into the small region.

4. The information processing device described in claim 3, wherein the control unit generates the segmented image based on the SAR image that has been subjected to threshold processing, and the threshold processing is a process of uniformly changing pixels that have a predetermined first luminance value or less that is darker than the luminance values ​​of the pixels representing the multiple vehicles to a second luminance value that is darker than the predetermined first luminance value.

5. An information processing device according to any one of claims 1 to 4, wherein the control unit obtains the number of segments, which is the number of small areas included in the divided image, and generates the divided image by dividing the SAR image according to the number of segments according to the SLIC algorithm.

6. The information processing device described in claim 5, wherein the control unit extracts one independent vehicle that is not included in the set from the SAR image, and obtains the number of segments based on the value obtained by dividing the number of pixels representing the independent vehicle from the total number of pixels in the SAR image.

7. The information processing device according to any one of claims 1 to 4, wherein the control unit generates the divided image in accordance with a watershed algorithm or a quickshift algorithm.

8. The information processing device according to any one of claims 1 to 7, wherein the control unit outputs the number of the plurality of vehicles included in the set based on the divided image.

9. An information processing method including a step of generating a segmented image in which a plurality of vehicles located close to each other are divided into separate small regions by grouping similar pixels from an SAR image containing the plurality of vehicles located close to each other into small regions of irregular shape.

10. A computer program that causes a computer to function as an information processing device, and that causes the computer to function as a control unit that generates a segmented image in which a group of multiple vehicles located close to each other are divided into separate small areas by grouping similar pixels from an SAR image containing the group of multiple vehicles located close to each other into small, irregular areas.

Images