Object tracking parameter setting support system, multi-camera object tracking system, object tracking parameter setting support program, and multi-camera object tracking program

Abstract

To perform accurate and high-speed multi-camera object tracking processing while suppressing economical and temporal cost by enabling automatic setting of a movable path in an object tracking parameter setting support system.SOLUTION: Multi-camera object tracking is performed relative to a photographed image from a plurality of fixed cameras without limitation of a movable path (S3), a movable path is estimated from statistical information of the tracking results (S5), and information on the movable path is set as a parameter for multi-camera object tracking. Thus, information on the movable path can be automatically set, so that economical and temporal cost can be suppressed. Further, since multi-camera object tracking with restriction of the movable path can be performed by using the information on the movable path (S7), accurate and high-speed multi-camera object tracking process can be achieved.SELECTED DRAWING: Figure 8

Description

Technical Field

[0001] The present invention relates to an object tracking parameter setting support system, a multi-camera object tracking system, an object tracking parameter setting support program, and a multi-camera object tracking program.

Background Art

[0002] Conventionally, a multi-camera object tracking technique has been known in which a plurality of cameras are installed in a certain area, and an object (such as a person) within the shooting ranges of these plurality of cameras is tracked using the captured images of these cameras (see, for example, Non-Patent Document 1). This multi-camera object tracking, to be precise, is to track a plurality of objects to be tracked across a plurality of cameras (captured images) while taking into account occlusion (which occurs when an object to be tracked is hidden from the camera, for example, when the object to be tracked enters the shadow of another object or moves in and out of the shooting range of the camera). By using the above multi-camera object tracking technique, it is possible to confirm the safety of an object to be tracked, analyze the behavior of the object to be tracked, and utilize the analysis results for reexamining the route and the arrangement of objects in the area where the camera is installed.

[0003] For a system that installs multiple cameras in a facility such as a store and checks the flow of individual people, like the above-described multi-camera object tracking system, it is important to add the physical restrictions of the building (facility) as information. For example, when going from a certain camera A to camera C, due to the physical restrictions of the building (facility), it is necessary to necessarily pass through the shooting range of camera B (for example, when there are obstacles on the path directly connecting camera A and camera C, etc.). Basically, the flow of people directly (linearly) going from camera A to camera C does not occur, and the data of the tracking results of people passing through such a path needs to be excluded. That is, in performing multi-camera object tracking, information about the path between cameras where people can move ("movable path") is important. This "movable path" is often represented by a graph. Specifically, each camera is often regarded as a node (vertex) in the graph, and the above-mentioned movable path (the path between cameras where people can move) is often regarded as an edge (a line connecting vertices) in the graph.

[0004] In order to ensure the accuracy and speed of the multi-camera object tracking process, it is very effective to utilize various spatial information including information about the above-mentioned movable path. This spatial information includes in-information (information about the in-area, which is the area where people enter, in the shooting range of each camera in the facility) and out-information (information about the out-area, which is the area where people go out, in the shooting range of each camera in the facility) of the shooting range of each camera in the facility, information about the entrance and exit areas of the facility, information about the area of interest (the area where the person to be tracked can pass through, or the area where the person to be tracked cannot pass through) in the shooting range of each camera in the facility, and mapping information between the shooting images of each camera in the facility and the map of the facility (the building floor plan) (information about the association between each point (each coordinate) in the shooting image (shooting range) of each camera and each point in the map of the facility).

Prior Art Documents

Non-Patent Documents

[0005]

Non-Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0006] As described above, in order to ensure the accuracy and processing speed of multi-camera object tracking, it is very effective to utilize various spatial information including information about movable paths. However, it takes economic and time costs to manually set the above spatial information.

[0007] The present invention solves the above problems, enables automatic setting of various spatial information such as movable paths necessary to ensure the accuracy and speed of multi-camera object tracking processing, saves the labor of manually setting these spatial information, suppresses economic and time costs, and realizes an accurate and high-speed multi-camera object tracking processing. An object of the present invention is to provide a parameter setting support system for object tracking, a multi-camera object tracking system, a parameter setting support program for object tracking, and a multi-camera object tracking program.

Means for Solving the Problems

[0008] To solve the above problems, a parameter setting support system for object tracking according to a first aspect of the present invention is a parameter setting support system for object tracking that obtains and sets parameters used for multi-camera object tracking. The system includes an input means for inputting captured images from a plurality of cameras, a multi-camera object tracking means for performing multi-camera object tracking to track a plurality of objects to be tracked across the shooting ranges of the plurality of cameras based on the captured images from the plurality of cameras, a movable path estimation means for performing multi-camera object tracking by the multi-camera object tracking means on the captured images of a predetermined period from the plurality of cameras and estimating a movable path, which is a path between cameras where a person can move, from the statistical information of the tracking results, and a setting means for setting the information of the movable path estimated by the movable path estimation means as one of the parameters used for the multi-camera object tracking.

[0009] In this parameter setting support system for object tracking, a single-camera object tracking means for performing single-camera object tracking to track one or more objects to be tracked within the shooting range of one of the plurality of cameras based on the captured image from one of the plurality of cameras, and an in / out area estimation means for performing single-camera object tracking by the single-camera object tracking means on the captured images of a predetermined period for each of the plurality of cameras and estimating an in-area, which is an area where a person enters, and an out-area, which is an area where a person exits, in the shooting range of each of the plurality of cameras from the statistical information of these tracking results are further provided. It is desirable that the setting means sets the information of the in-area and the out-area estimated by the in / out area estimation means as one of the parameters used for the multi-camera object tracking.

[0010] In this parameter setting support system for object tracking, for the captured images of a predetermined period from the plurality of cameras, multi-camera object tracking is performed by the multi-camera object tracking means, and from the statistical information of this tracking result, an entrance / exit estimation means for estimating the entrance and exit of the facility where the plurality of cameras are arranged is further provided. It is desirable that the setting means sets the information on the entrance and exit estimated by the entrance / exit estimation means as one of the parameters used for the multi-camera object tracking.

[0011] In this parameter setting support system for object tracking, mapping means for associating each point in the captured image of each of the plurality of cameras with each point in the map of the facility where the plurality of cameras are arranged is further provided. It is desirable that the setting means sets the information on the association result by the mapping means as one of the parameters used for the multi-camera object tracking.

[0012] In this parameter setting support system for object tracking, the movable path estimation means may perform multi-camera object tracking by the multi-camera object tracking means on the captured images of a predetermined period from the plurality of cameras without restricting the movable path, which is the path between the cameras where a person can move, and as a result of this tracking, remove the paths between the cameras with fewer passing people from the candidates for the movable path to estimate the movable path.

[0013] The multi-camera object tracking system according to the second aspect of the present invention is a multi-camera object tracking system that performs multi-camera object tracking on an object reflected in captured images from a plurality of cameras. The multi-camera object tracking system includes multi-camera object tracking means for tracking a plurality of objects to be tracked across the imaging ranges of the plurality of cameras based on the captured images from the plurality of cameras. Using the information on the movable path set by the above object tracking parameter setting support system, the multi-camera object tracking means performs multi-camera object tracking with restrictions on the movable path on the captured images from the plurality of cameras.

[0014] The object tracking system according to the third aspect of the present invention includes single-camera object tracking means for performing single-camera object tracking to track one or more objects to be tracked within the imaging range of one camera based on a captured image from the one camera, and interest area estimation means for performing single-camera object tracking by the single-camera object tracking means on the captured images of the one camera for a predetermined period, and estimating, from the statistical information of the tracking results, an area of interest that is an area through which an object to be tracked can pass or an area through which an object to be tracked cannot pass in the imaging range of the one camera.

[0015] The parameter setting support program for object tracking according to the fourth aspect of the present invention is a parameter setting support program for object tracking for obtaining and setting parameters used for multi-camera object tracking. The program causes a computer to function as: input means for inputting captured images from a plurality of cameras; multi-camera object tracking means for performing multi-camera object tracking to track a plurality of objects to be tracked across the shooting ranges of the plurality of cameras based on the captured images from the plurality of cameras; movable path estimation means for performing multi-camera object tracking by the multi-camera object tracking means on the captured images for a predetermined period from the plurality of cameras and estimating a movable path, which is a path between cameras where a person can move, from statistical information of the tracking results; and setting means for setting information on the movable path estimated by the movable path estimation means as one of the parameters used for the multi-camera object tracking.

[0016] The multi-camera object tracking program according to the fifth aspect of the present invention is a multi-camera object tracking program for performing multi-camera object tracking on an object captured in a captured image from a plurality of cameras. The program causes a computer to function as multi-camera object tracking means for performing multi-camera object tracking to track a plurality of objects to be tracked across the shooting ranges of the plurality of cameras based on the captured images from the plurality of cameras, and perform multi-camera object tracking with restrictions on the movable path on the captured images from the plurality of cameras by the multi-camera object tracking means using the information on the movable path set using the above-described parameter setting support program for object tracking.

[0017] The object tracking program according to the sixth aspect of the present invention is an object tracking program for performing object tracking on an object reflected in a captured image from one camera, the computer based on the captured image from the one camera, single camera object tracking means for tracking one or more objects to be tracked within the shooting range of the one camera, and for the captured images of the one camera for a predetermined period, single camera object tracking is performed by the single camera object tracking means, and from the statistical information of this tracking result, a region where an object to be tracked can pass or a region of interest that is a region where an object to be tracked cannot pass in the shooting range of the one camera is estimated.

Effect of the Invention

[0018] According to the object tracking parameter setting support system according to the first aspect of the present invention and the object tracking parameter setting support program according to the fourth aspect, by inputting captured images from a plurality of cameras, based on the captured images from the plurality of input cameras, a movable path is estimated, and the information of the estimated movable path can be set as one of the parameters used for multi-camera object tracking. As a result, the information on the movable path required for performing multi-camera object tracking with restrictions on the movable path can be automatically set, saving the labor of manually setting the information on the movable path and suppressing economic and time costs. Further, by using the above information on the movable path, multi-camera object tracking with restrictions on the movable path can be performed on the captured images from a plurality of cameras. As a result, compared with the case of performing multi-camera object tracking without restrictions on the movable path, the path (the path between cameras) can be narrowed down (limited) to a path where a person can move, and multi-camera object tracking can be performed, so that an accurate and high-speed multi-camera object tracking process can be realized.

[0019] Also, according to the multi-camera object tracking system according to the second aspect of the present invention and the multi-camera object tracking program according to the fifth aspect, using the information on the movable path set using the object tracking parameter setting support system according to the above first aspect or the object tracking parameter setting support program according to the fourth aspect, the multi-camera object tracking means performs multi-camera object tracking with restrictions on the movable path for the captured images from a plurality of cameras. Thereby, the labor of manually setting the information on the movable path can be saved, and economic and time costs can be suppressed. Further, as described above, by using the information on the movable path to perform multi-camera object tracking with restrictions on the movable path for the captured images from a plurality of cameras, compared with the case of performing multi-camera object tracking without restrictions on the movable path, the path (the route between cameras) is narrowed down (limited) to a path where a person can move, so that multi-camera object tracking can be performed, and accurate and high-speed multi-camera object tracking processing can be realized.

[0020] Also, according to the object tracking system according to the third aspect of the present invention and the object tracking program according to the sixth aspect, single-camera object tracking is performed on the captured images of one camera for a predetermined period, and from the statistical information of the tracking results, a region where the object to be tracked can pass or a region of interest (ROI) that is a region where the object to be tracked cannot pass in the shooting range of the above one camera is estimated. By using the information on the above region of interest, object tracking can be performed by limiting the region where the person to be tracked can pass with reference to the information on the above region of interest, so that more accurate and high-speed object tracking processing can be realized.

Brief Description of the Drawings

[0021]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Embodiments for Carrying Out the Invention

[0022] Hereinafter, a parameter setting support system for object tracking, a multi-camera object tracking system, a parameter setting support program for object tracking, and a multi-camera object tracking program according to an embodiment embodying the present invention will be described with reference to the drawings. FIG. 1 is a block configuration diagram showing a schematic configuration of a multi-camera object tracking system 10 according to the present embodiment. As shown in FIG. 1, in the present embodiment, a plurality of fixed cameras 3 (corresponding to the "camera" in the claims), which are network cameras for monitoring that photograph a predetermined photographing area, and an edge-side analysis device 2 that analyzes the video from each fixed camera 3 are arranged in a store S (corresponding to the "facility" in the claims) such as a chain store. An example of this case will be described.

[0023] The above multi-camera object tracking system 10 mainly includes the above plurality of fixed cameras 3 and edge-side analysis devices 2 installed in each store S, and an analysis server 1 (corresponding to the "computer" in the claims) arranged on the cloud C. The edge-side analysis device 2 and the analysis server 1 are connected via the Internet.

[0024] As shown in FIG. 1, the above multi-camera object tracking system 10 includes a hub 7 and a router 8 in the store S where the above fixed cameras 3 and edge-side analysis devices 2 are arranged. In the following description, an example where the fixed camera 3 and the edge-side analysis device 2 are separate will be described. However, the camera in the multi-camera object tracking system of the present invention may be a so-called AI camera in which the above fixed camera 3 and edge-side analysis device 2 are integrated.

[0025] The above-mentioned fixed camera 3 has an IP address and can be directly connected to the network. As shown in FIG. 1, the edge-side analysis device 2 is connected to a plurality of fixed cameras 3 via a LAN (Local Area Network) and a hub 7, and analyzes the images input from each of these fixed cameras 3. More specifically, the edge-side analysis device 2 performs object detection processing (specifically, detection processing of a person (bounding box)) on the images input from each of the fixed cameras 3, and cuts out the images of the persons detected by this object detection processing, and extracts feature vectors from the cut-out images of the persons.

[0026] The above-mentioned analysis server 1 is a server installed in a management department (such as the head office) that comprehensively manages each store S. Although details will be described later, the analysis server 1 uses the information of the bounding box (position and size) of each person (each object) and the feature vector of each person transmitted from the edge-side analysis device 2 installed in each store S, and performs multi-camera object tracking processing, which is a process of tracking a plurality of persons to be tracked across the shooting ranges of a plurality of cameras while considering occlusion (a state where an object in front hides an object behind so that it cannot be seen). In addition, the analysis server 1 includes an object tracking parameter setting support system 40 (see FIG. 5), and based on the input shooting images from a plurality of fixed cameras 3 and the data of a two-dimensional store map, obtains and sets parameters of various spatial information used for the above-mentioned multi-camera object tracking processing.

[0027] Next, with reference to FIG. 2, the hardware configuration of the edge-side analyzer 2 will be described. The edge-side analyzer 2 includes a System-on-a-Chip (SoC) 21, a hard disk 22 for storing various data and programs, a Random Access Memory (RAM) 23, and a communication control IC 25. The SoC 21 includes a CPU 21a that controls the entire device and performs various calculations, and a GPU 21b that is used for inference processing of various learned DNN (Deep Neural Networks) models including the above-mentioned object detection processing and feature vector extraction processing. Further, the data stored in the hard disk 22 includes video data after decoding the video stream (data) input from each of the fixed cameras 3. Further, the programs stored in the hard disk 22 include various learned DNN models for various inference processes (various learned DNN models for various inference processes) including the above-mentioned object detection processing and feature vector extraction processing, and an analyzer control program.

[0028] Next, with reference to FIG. 3, the hardware configuration of the analysis server 1 will be described. The analysis server 1 includes a CPU 31 that controls the entire device and performs various calculations, a hard disk 32 for storing various data and programs, a Random Access Memory (RAM) 33, a display 34, an operation unit 35, and a communication unit 36. The programs stored in the hard disk 32 include an object tracking parameter setting support program 37 and a multi-camera object tracking program 38.

[0029] Next, an overview of the above multi-camera object tracking process will be described. This multi-camera object tracking process is a process of tracking a plurality of objects to be tracked (humans in this embodiment) across the shooting ranges of a plurality of cameras (fixed cameras 3 in this embodiment) while taking occlusion into consideration. As shown in FIG. 4, this process repeatedly performs a process of determining whether the humans shown in the captured images of different fixed cameras 3 are the same person, using the captured images of a plurality of fixed cameras 3 (three fixed cameras 3a to 3c in the example of FIG. 4), thereby tracking a plurality of people across the shooting ranges of the plurality of fixed cameras 3. Through this tracking process, it is possible to know what routes each person has taken in the store S.

[0030] The process of determining whether the humans shown in the captured images of the different fixed cameras 3 are the same person is performed as follows, for example, considering the consistency between the time when the object crosses cameras and the current position of the object. As shown in query X in FIG. 4, for the humans (Person 1 and Person 2) shown in the captured image of fixed camera 3b at 15:00, matching is performed only with the humans who passed through the shooting range of fixed camera 3a (appeared in the captured image of fixed camera 3a) within 2 minutes (from 14:58 to 15:00). Thereby, it can be known that Person 1 moved from the shooting range of fixed camera 3a to the shooting range of fixed camera 3b. Also, as shown in query Y in FIG. 4, for the human (Person 3) shown in the captured image of fixed camera 3c at 16:00, matching is performed only with the humans who passed through the shooting range of fixed camera 3a (appeared in the captured image of fixed camera 3a) within 10 minutes (from 15:50 to 16:00) and the humans who passed through fixed camera 3b (appeared in the captured image of fixed camera 3b) within 2 minutes (from 14:58 to 15:00). Thereby, it can be known that Person 3 moved from the shooting range of fixed camera 3a to the shooting range of fixed camera 3c.

[0031] Next, with reference to FIG. 5, an overview of the parameter setting process for multi-camera object tracking performed by the above-described analysis server 1 will be described. FIG. 5 shows the input / output relationships of blocks related to the (automatic) setting process of parameters (various spatial information) used for multi-camera object tracking (i.e., blocks constituting the object tracking parameter setting support system 40) and blocks related to the multi-camera object tracking process with path restrictions among the functional blocks of the CPU 31 of the analysis server 1. The input unit 41, the mapping unit 43, the single-camera object tracking unit 44, the in / out area estimation unit 45, the region of interest estimation unit 46, the edge information estimation unit 48, the entrance / exit estimation unit 49, and the setting unit 50 in FIG. 5 correspond to the input means, the mapping means, the single-camera object tracking means, the in / out area estimation means, the region of interest estimation means, the movable path estimation means, the entrance / exit estimation means, and the setting means in the claims, respectively. Also, the multi-camera object tracking unit 47 without path restrictions and the multi-camera object tracking unit 51 with path restrictions in FIG. 5 correspond to the multi-camera object tracking means in the claims.

[0032] The above-described input unit 41 inputs the captured images from each of the plurality of fixed cameras 3 (for example, fixed cameras 3a to 3c in FIG. 4) and the data of the two-dimensional map of the store (hereinafter referred to as "two-dimensional store map") where the above-described plurality of fixed cameras 3 are arranged. The above-described two-dimensional store map is, for example, a store map (two-dimensional store map 60) as shown in FIG. 6. Note that, in FIG. 6, the two-dimensional store map 60 is shown in a simplified manner for convenience of explanation.

[0033] The mapping unit 43 shown in FIG. 5 performs the association between each point (each coordinate) in the captured image of each of the plurality of fixed cameras 3 and each point in the above-described two-dimensional store map 60.

[0034] In addition, the single camera object tracking unit 44 performs single camera object tracking, which is a process of tracking one or more persons to be tracked within the shooting range of one of the plurality of fixed cameras 3 (for example, the fixed camera 3a) based on a captured image from the one camera. The in / out area estimation unit 45 performs single camera object tracking by the single camera object tracking unit 44 on the captured images of each of the plurality of fixed cameras 3 for a predetermined period, and estimates, from the statistical information of these tracking results, an in-area, which is an area where people enter, and an out-area, which is an area where people go out, in the shooting range of each of the plurality of fixed cameras 3. Further, the region of interest estimation unit 46 performs single camera object tracking by the single camera object tracking unit 44 on the captured images of each of the plurality of fixed cameras 3 for a predetermined period, and estimates, from the statistical information of these tracking results, a region of interest, which is an area where the person to be tracked can pass or an area where the person to be tracked cannot pass, in the shooting range of each of the plurality of fixed cameras 3.

[0035] In addition, the multi-camera object tracking unit 47 without path restriction in FIG. 5 performs multi-camera object tracking to track a plurality of persons to be tracked across the shooting ranges of a plurality of fixed cameras 3 for the captured images of a predetermined period from the plurality of fixed cameras 3 without restricting the movable path. Here, the above-mentioned "movable path" means the path between the fixed cameras 3 where a person can move in the two-dimensional store map 60. Further, the edge information estimation unit 48 estimates the above-mentioned movable path from the statistical information of the multi-camera object tracking result by the multi-camera object tracking unit 47 without path restriction. Note that the naming of the above-mentioned edge information estimation unit 48 is derived from the fact that each edge in the graph representing the path (pass) between the fixed cameras 3 where a person can move in the graph information expressing the relationship between the fixed cameras 3, which will be described later, is regarded as the above-mentioned movable path. Note that each node in the graph representing the path (pass) between the fixed cameras 3 where a person can move is each fixed camera 3. 55, 56, and 57 in FIG. 6 represent a path in the AB direction (that is, a path between the fixed camera 3a and the fixed camera 3c), a path in the BC direction (that is, a path between the fixed camera 3b and the fixed camera 3c), and a path in the AC direction (that is, a path between the fixed camera 3a and the fixed camera 3c), respectively.

[0036] Further, the entrance / exit estimation unit 49 estimates an entrance area 58 and an exit area 59 (see FIG. 6) of the store S where the plurality of fixed cameras 3 are arranged, from the statistical information of the multi-camera object tracking result by the above-described multi-camera object tracking unit 47 without path restriction. Then, the setting unit 50 sets, as parameters used for multi-camera object tracking, the in-area and out-area estimated by the in / out area estimation unit 45, the region of interest estimated by the region of interest estimation unit 46, the movable path estimated by the edge information estimation unit 48, the entrance and exit information (entrance area 58 and exit area 59) estimated by the entrance / exit estimation unit 49, and the information of the association result (the association result between each point in the captured image of each fixed camera 3 and each point in the two-dimensional store map 60) performed by the mapping unit 43. Note that, for the multi-camera object tracking by the multi-camera object tracking unit 47 without path restriction, the information of the single-camera object tracking result by the single-camera object tracking unit 44 is used. Therefore, as described above, when both the in-area / out-area and the region of interest estimated from the statistical information of the single-camera object tracking result and the movable path and the entrance and exit information (entrance area 58 and exit area 59) estimated from the statistical information of the multi-camera object tracking result without path restriction are set as parameters used for multi-camera object tracking, it is desirable to connect the multi-camera object tracking unit 47 without path restriction in series after the single-camera object tracking unit 44, so as to reduce the overall computational amount of the object tracking parameter setting support system 40.

[0037] Among the functional blocks of the CPU 31 of the analysis server 1 in FIG. 5, an input unit 41, a mapping unit 43, a single camera object tracking unit 44, an in / out area estimation unit 45, an area of interest estimation unit 46, a multi-camera object tracking unit 47 without path restriction, an edge information estimation unit 48, an entrance / exit estimation unit 49, and a setting unit 50 constitute an object tracking parameter setting support system 40. Further, the multi-camera object tracking unit 51 with path restriction in FIG. 5 performs multi-camera object tracking that tracks a plurality of persons to be tracked across the shooting ranges of a plurality of fixed cameras 3 with restrictions on movable paths, based on the shooting images from the plurality of fixed cameras 3, based on various parameters for multi-camera object tracking set by the setting unit 50 of the above object tracking parameter setting support system 40. Note that the multi-camera object tracking process by the above multi-camera object tracking unit 51 with path restriction is the actual multi-camera object tracking process performed by the multi-camera object tracking system 10 based on various parameters for multi-camera object tracking set in advance by the object tracking parameter setting support system 40.

[0038] In order to ensure the accuracy and speed of the above actual multi-camera object tracking process, it is very effective to utilize various spatial information including information about the above movable paths. In particular, in addition to the information about the above movable paths, in information (information about the in area, which is the area where people enter, in the shooting range of each fixed camera 3) and out information (information about the out area, which is the area where people go out, in the shooting range of each fixed camera 3) of the shooting ranges of the respective fixed cameras 3, and information about the entrance and exit areas of the store S (see the entrance area 58 and the exit area 59 in FIG. 6) are very important. After setting these spatial information in advance, by performing the actual multi-camera object tracking process with path restrictions, an accurate and high-speed multi-camera object tracking process can be realized.

[0039] As described above, in order to ensure the accuracy and processing speed of multi-camera object tracking, it is very effective to utilize various spatial information including information about the movable path. However, it takes economic and time costs to manually set the above spatial information.

[0040] Therefore, in the object tracking parameter setting support system 40 of the present embodiment, based on the captured images from a plurality of fixed cameras 3 and the data of the two-dimensional store map 60, the above various spatial information (including information about the movable path) is automatically obtained, and these spatial information are described in the graph information (see FIG. 7) representing the relationship between the fixed cameras 3 (that is, the various spatial information are automatically graph-structured and represented). Then, by using this graph information to perform the multi-camera object tracking process with path restrictions in the actual operation (using the graph information as the input of the multi-camera object tracking process), the labor of manually setting the above spatial information can be saved, and the economic and time costs can be suppressed. Further, by using the above movable path, in-information and out-information, information on the entrance and exit areas, and information on the area of interest, the path between cameras where a person can move, the area where a person enters and the area where a person exits in the shooting range of each fixed camera 3 (it may be the direction where a person enters and the direction where a person exits), the entrance and exit of the store S, and the area where the person to be tracked can pass (or the area where the person to be tracked cannot pass) are narrowed down, and by performing the multi-camera object tracking process with path restrictions in the actual operation, an accurate and high-speed multi-camera object tracking process can be performed.

[0041] Next, in addition to the flowchart of FIG. 8, referring to FIGS. 5, 6, 7, 9, 10, and 11, the processing performed by the multi-camera object tracking system 10 of the present embodiment will be described. Note that the processing of S1 to S6 in the flowchart of FIG. 8 is the processing performed by the object tracking parameter setting support system 40. First, the input unit 41 (see FIG. 5) of the analysis server 1 inputs the captured images from each fixed camera 3 and the data of the two-dimensional store map 60 (see FIGS. 6 and 7) (S1).

[0042] Next, the mapping unit 43 of the analysis server 1 performs an association (mapping) between each point (each coordinate) in the captured image of each fixed camera 3 and each point in the two-dimensional store map 60 (S2). More specifically, based on the installation location and height of each fixed camera 3, the mapping unit 43 uses mathematical conversion to associate each point (for example, each coordinate of the captured image of a 360-degree camera (fisheye camera)) in the captured image of each fixed camera 3 arranged in the store with each point in the two-dimensional store map 60.

[0043] Next, referring to the association (mapping) result obtained in S2 above, the multi-camera object tracking unit 47 without path restriction and the single-camera object tracking unit 44 of the analysis server 1 perform multi-camera object tracking without movable path restriction, including single-camera object tracking, on the captured images (videos for a certain period) from a plurality of fixed cameras 3 for a predetermined period (S3).

[0044] Next, the in / out area estimation unit 45 and the region of interest estimation unit 46 of the analysis server 1 estimate (limit) the in-area, out-area, and region of interest in the shooting range of each fixed camera 3 from the statistical information of the single-camera object tracking result by the single-camera object tracking unit 44 (S4).

[0045] First, the estimation process of the above-mentioned in-area and out-area will be described. For example, the in / out area estimation unit 45 of the analysis server 1 estimates, as shown in FIG. 11, an in-area 62, which is an area where people enter, and an out-area 63, which is an area where people leave, within the shooting range 61 of each fixed camera 3, from the statistical information of the single camera object tracking result by the single camera object tracking unit 44. Then, the in / out area estimation unit 45 estimates, for each node (for example, Node A (fixed camera 3a)) shown in FIG. 7, an in-direction (the direction in which people enter (for example, "left", "right", "down")) and an out-direction (the direction in which people leave (for example, "right", "down")) from the estimation results of the above-mentioned in-area and out-area.

[0046] Next, the estimation process of the region of interest in the above-mentioned S4 will be described. The region of interest estimation unit 46 of the analysis server 1 estimates a region of interest (ROI (Region of Interest)), which is an area where the person to be tracked can pass or an area where the person to be tracked cannot pass, within the shooting range 61 of each fixed camera 3, from the statistical information of the single camera object tracking result by the single camera object tracking unit 44. The area where the person to be tracked cannot pass is, for example, an area where there are obstacles (such as product shelves) in the shooting range of each fixed camera 3. When the shape of the ROI (region of interest) is square or rectangular, as shown in FIG. 7, the ROI is defined by its width (horizontal length), height (vertical length), and the x and y coordinates of the origin. Note that, as described above, when the shape of the ROI is square or rectangular, the origin is, for example, the upper left corner (coordinates) of the ROI.

[0047] Next, the edge information estimation unit 48 of the analysis server 1 estimates the above-mentioned movable path (edge information), the entrance area 58, and the exit area 59 of the store S from the statistical information of the multi-camera object tracking result without path restrictions by the above-mentioned multi-camera object tracking unit 47 without path restrictions (S5).

[0048] First, the above-described estimation process of the movable path will be explained. For example, the edge information estimation unit 48 of the analysis server 1 estimates (determines) the above-described movable path by removing (excluding) paths (routes) below the threshold value in the above-described statistical information 81 of the tracking result (see FIG. 9). More specifically, the edge information estimation unit 48 removes a path between fixed cameras 3 with few passing people from the candidates for the movable path as a result of multi-camera object tracking without path restrictions by the above-described multi-camera object tracking unit 47 without path restrictions, thereby estimating the above-described movable path. For example, in the case of the example shown in FIG. 9, since the threshold value is set to 10, the edge information estimation unit 48 removes the paths A->C (the path from the captured image A to the captured image C (that is, the path from the fixed camera 3a to the fixed camera 3c)), B->D (the path from the captured image B to the captured image D (that is, the path from the fixed camera 3b to the fixed camera 3d)), C->A (the path from the captured image C to the image A (that is, the path from the fixed camera 3c to the fixed camera 3a)), and D->B (the path from the captured image D to the captured image B (that is, the path from the fixed camera 3d to the fixed camera 3b)) from the candidates for the movable path, and estimates that the remaining paths (A->B, A->D, B->A, B->C, C->B, C->D, D->A, and D->C) are the movable paths. That is, the edge information estimation unit 48 estimates that the edge information (movable path) in the two-dimensional store map 60 of FIG. 9 is A<=>B, B<=>C, C<=>D, A<=>D.

[0049] In the example shown in FIG. 9, the edge information (movable path) related to Node A (fixed camera 3a) is presumed to be A<=>B and A<=>D. In the case of this example, since there is an obstacle called a shelf 82 in A<=>C (the path connecting captured image A and captured image C (that is, the path connecting fixed camera 3a and fixed camera 3c)), a flow of people directly (linearly) from fixed camera 3a to fixed camera 3c does not occur, and data on tracking results of (people) passing through such a path needs to be excluded. For this reason, the edge information estimation unit 48 removes A<=>C (the path from A to C and the path from C to A) from the candidates for the edge information (movable path) of Node A (fixed camera 3a).

[0050] Next, the estimation process of the entrance area 58 and the exit area 59 (see FIG. 10) in S5 above will be described. The entrance / exit estimation unit 49 of the analysis server 1 estimates the entrance area 58 and the exit area 59 (see FIG. 10) of the store S where a plurality of fixed cameras 3 are arranged from the statistical information of the multi-camera object tracking results by the multi-camera object tracking unit 47 without the above path restriction. Then, from these estimation results, entrance information and exit information for each node (for example, Node A (fixed camera 3a)) (whether there are an entrance and an exit within the shooting range of each fixed camera 3 and their directions) are estimated.

[0051] For example, the entrance / exit estimation unit 49 of the analysis server 1 estimates (determines) the above-described entrance area 58 and exit area 59 by removing (excluding) routes below a threshold value from among the routes in the statistical information 83 (see FIG. 10) of the multi-camera object tracking result by the above-described multi-camera object tracking unit 47 without path restrictions. More specifically, the entrance / exit estimation unit 49 removes, from candidates for the route (hereinafter referred to as the "entrance / exit route") from the entrance area 58 to the exit area 59, the routes between three fixed cameras with few people passing through, which are the results of multi-camera object tracking without path restrictions by the above-described multi-camera object tracking unit 47 without path restrictions, thereby narrowing down the above-described entrance / exit route. Then, the in-area 62 with many people entering and the out-area 63 (see FIG. 11) with many people leaving in the shooting range 61 of each fixed camera 3 corresponding to the start point and the end point in the narrowed-down entrance / exit route are estimated (determined) as the entrance area 58 and the exit area 59. For example, in the case of the example shown in FIG. 10, since the threshold value is set to 10, the entrance / exit estimation unit 49 removes BCD (the route from fixed camera 3b through fixed camera 3c to fixed camera 3d), BAC (the route from fixed camera 3b through fixed camera 3a to fixed camera 3c), and CBA (the route from fixed camera 3c through fixed camera 3b to fixed camera 3a) from the candidates for the above-described entrance / exit route, thereby narrowing down the entrance / exit route to ABC, ADC, ABADC, and ADABC. Each fixed camera 3 corresponding to the start point and the end point in each of the above-described narrowed-down entrance / exit routes (ABC, ADC, ABADC, and ADABC) is the fixed camera 3a and the fixed camera 3c. The entrance / exit estimation unit 49 uses the statistical information of the above-described single-camera object tracking result to estimate (determine) the in-area 62 (see FIG. 11) with many people entering in the shooting range 61a of the start point (fixed camera 3a) of the above-described narrowed-down entrance / exit route as the entrance area 58, and estimates (determines) the out-area 63 (see FIG. 11) with many people leaving in the shooting range 61c of the end point (fixed camera 3c) of the above-described narrowed-down entrance / exit route as the exit area 59.

[0052] Next, the setting unit 50 of the analysis server 1 automatically constructs graph information 80 for input to multi-camera object tracking by using the information on the estimation results from the processes in S4 and S5 (inward and outward directions, ROI (region of interest), edge information (movable path), and entrance information and exit information (entrance area and exit area)), and sets (outputs) the graph information 80 and the information on the association result (hereinafter referred to as "mapping information") between each point in the captured image of each fixed camera 3 estimated in S2 and each point in the two-dimensional store map 60 as parameters to be used for multi-camera object tracking (S6).

[0053] Then, the multi-camera object tracking unit 51 with path restriction in the analysis server 1 (see FIG. 5) uses the above graph information 80 and mapping information as input parameters to execute multi-camera object tracking with restrictions on the movable path (S7).

[0054] Here, with reference to FIG. 7, a specific example of the above graph information 80 will be described in detail. For example, the fixed cameras 3 arranged in the store S are three cameras, namely fixed cameras 3a, 3b, and 3c, and they are arranged as shown in the two-dimensional store map 60 on the left side of FIG. 7. Assume that there is an obstacle, a shelf 82, on the path connecting the fixed camera 3a and the fixed camera 3c. In this case, the graph information 80 is composed of the graph information about Node A (fixed camera 3a), the graph information about Node B (fixed camera 3b), and the graph information about Node C (fixed camera 3c). And in the example shown in FIG. 7, in the graph information about Node A (fixed camera 3a), the edge information (movable path) is A<=>B, the inward direction is left, right, and down, the outward direction is right and down, the entrance information is "Entrance Yes (left)" (information indicating that there is an entrance within the shooting range of the fixed camera 3a and the entrance is on the left side), the exit information is "Exit No" (information indicating that there is no exit within the shooting range of the fixed camera 3a), and the ROI (region of interest) is information about the region defined by its width, height, and the x and y coordinates of the origin.

[0055] As described above, according to the parameter setting support system 40 for object tracking of the present embodiment, by inputting the captured images from the plurality of fixed cameras 3 and the data of the map (two-dimensional store map 60) of the store S where these fixed cameras 3 are arranged, based on the input captured images from the plurality of fixed cameras 3 and the data of the two-dimensional store map 60, a movable path (edge information) is estimated, and the information of the estimated movable path can be set as one of the parameters used for multi-camera object tracking. Thereby, the information of the movable path necessary for performing multi-camera object tracking with restrictions on the movable path can be automatically set, so that the labor of manually setting the information of the movable path can be saved, and economic and time costs can be reduced. Further, by using the above-mentioned information of the movable path, multi-camera object tracking with restrictions on the movable path can be performed on the captured images from the plurality of fixed cameras 3. Thereby, compared with the case of performing multi-camera object tracking without restrictions on the movable path, the path (the route between fixed cameras) can be narrowed down (limited) to a path where a person can move, and multi-camera object tracking can be performed, so that accurate and high-speed multi-camera object tracking processing can be realized.

[0056] Moreover, according to the object tracking parameter setting support system 40 of the present embodiment, for the captured images of each fixed camera 3 during a predetermined period, single camera object tracking is performed, and from the statistical information of these tracking results, in the shooting range 61 of each fixed camera 3, an in-area 62 which is an area where people enter and an out-area 63 which is an area where people go out are estimated, and the information of the in-area 62 and the out-area 63 is set as one of the parameters used for multi-camera object tracking. Thereby, in the shooting range 61, the in-area 62 which is an area where people enter and the out-area 63 (in the example shown in the graph information 80 of FIG. 7, the direction where people enter (in-direction) and the direction where people go out (out-direction) in the shooting range) can be automatically set, so the trouble of manually setting the information of the in-area 62 and the out-area 63 (or the direction where people enter (in-direction) and the direction where people go out (out-direction)) can be saved, and economic and time costs can be suppressed. Further, by using the information of the in-area 62 and the out-area 63 (or the direction where people enter and the direction where people go out) described above, the in-area 62 and the out-area 63 (or the direction where people enter and the direction where people go out) in the shooting range 61 can be limited, and multi-camera object tracking can be performed, so a more accurate and high-speed multi-camera object tracking process can be realized.

[0057] Moreover, according to the parameter setting support system 40 for object tracking of the present embodiment, for the captured images of a predetermined period from a plurality of fixed cameras 3, multi-camera object tracking without movable path restriction is performed, and from the statistical information of this tracking result, the entrance information and the exit information of the store S are estimated, and the estimated entrance information and exit information are set as one of the parameters used for multi-camera object tracking. Thereby, since the entrance information and the exit information of the store S can be automatically set, the labor of manually setting the entrance information and the exit information can be saved, and economic and time costs can be suppressed. Further, by using the above-mentioned entrance information and exit information, the entrances and exits of the store S are limited, and multi-camera object tracking can be performed, so that a more accurate and high-speed multi-camera object tracking process can be realized.

[0058] Moreover, according to the parameter setting support system 40 for object tracking of the present embodiment, for the captured images of a predetermined period of each fixed camera 3, single-camera object tracking is performed, and from the statistical information of these tracking results, in the shooting range 61 of each fixed camera 3, a region where a person to be tracked can pass or a region of interest (ROI) that is a region where a person to be tracked cannot pass is estimated, and the information of the estimated region of interest is set as one of the parameters used for the multi-camera object tracking. Thereby, since the information of the above-mentioned region of interest can be automatically set, the labor of manually setting the information of the region of interest can be saved, and economic and time costs can be suppressed. Further, by using the information of the above-mentioned region of interest, with reference to the information of the above-mentioned region of interest, the region where a person to be tracked can pass is limited, and multi-camera object tracking can be performed, so that a more accurate and high-speed multi-camera object tracking process can be realized.

[0059] Further, according to the parameter setting support system 40 for object tracking of the present embodiment, the association (mapping) between each point (each coordinate) in the captured image of each fixed camera 3 and each point in the two-dimensional store map 60 is performed, and the information of the association result (mapping information) is set as one of the parameters used for multi-camera object tracking. Thereby, since the information of the above-mentioned association result (mapping information) can be automatically set, the labor of manually setting the mapping information can be saved, and the economic and time costs can be suppressed.

[0060] Further, according to the parameter setting support system 40 for object tracking of the present embodiment, multi-camera object tracking without movable path restriction is performed on the captured images for a predetermined period from a plurality of fixed cameras 3, and as a result of this tracking, the path between cameras with few passing people is removed from the candidates for movable paths, thereby estimating the movable path. Thereby, the information of the movable path necessary for performing multi-camera object tracking with a restricted movable path can be easily and automatically set.

[0061] Moreover, according to the multi-camera object tracking system 10 of the present embodiment, multi-camera object tracking with restrictions on the movable path is performed on the captured images from the plurality of fixed cameras 3 by using the information on the movable path set by the above-described object tracking parameter setting support system 40. Thereby, the labor of manually setting the information on the movable path can be saved, and economic and time costs can be reduced. Further, as described above, by performing multi-camera object tracking with restrictions on the movable path on the captured images from the plurality of fixed cameras 3 by using the information on the movable path, compared with the case of performing multi-camera object tracking without restrictions on the movable path, the path (the route between the fixed cameras) can be narrowed down (limited) to a path where a person can move, and multi-camera object tracking can be performed, so that accurate and high-speed multi-camera object tracking processing can be realized.

[0062] Modification example: Note that the present invention is not limited to the configurations of the above-described embodiments, and various modifications are possible without changing the gist of the invention. Next, modification examples of the present invention will be described.

[0063] Modification example 1: In the above embodiment, an example in which the object tracking parameter setting support system 40 automatically sets all of the edge information (movable path), entrance information and exit information, in-direction and out-direction, ROI (region of interest), and the above-described mapping information among the various spatial information parameters necessary to ensure the accuracy and processing speed of multi-camera object tracking has been described. However, the object tracking parameter setting support system of the present invention is not limited to this. For example, among the above-described various spatial information parameters, some parameters (for example, entrance information and exit information, in-direction and out-direction) may be input by the user of the analysis server using the operation unit in advance, or spatial information parameters transmitted from other computers may be used.

[0064] Modification Example 2: In the above embodiment, an example where the analysis server 1 performs the processing of all the functional blocks of the object tracking parameter setting support system 40 in FIG. 5 has been shown. However, the object tracking parameter setting support system of the present invention is not limited to this. For example, the edge-side analysis device may perform the processing of the mapping unit, the single-camera object tracking unit, the in / out area estimation unit, and the region of interest estimation unit in FIG. 5, and the analysis server may use the mapping information (information on the association result between each point of the captured image of each fixed camera and each point in the two-dimensional store map) transmitted from the edge-side analysis device, the information on the in-area and out-area (in-direction and out-direction), and the ROI (region of interest) as parameters for multi-camera object tracking.

[0065] Modification Example 3: In the above embodiment, an example where the edge-side device is a combination of the fixed camera 3 and the edge-side analysis device 2 has been shown. However, the edge-side device in the object tracking parameter setting support system of the present invention is not limited to this. For example, it may be a so-called AI camera, or a combination of a signage built into the camera and the edge-side analysis device.

[0066] Modification Example 4: In the above embodiment, from the statistical information of the single-camera object tracking results, as shown in FIG. 11, in the shooting range 61 of each fixed camera 3, the in-area 62 which is the area where people enter and the out-area 63 which is the area where people go out are estimated. From the estimation results of the above in-area and out-area, the in-direction and the out-direction for each node are estimated. That is, in the above embodiment, an example is shown where the information of the in-area and the out-area in the object tracking parameter setting support system of the present invention is the in-direction and the out-direction. However, the information of the in-area and the out-area in the object tracking parameter setting support system of the present invention is not limited to this, and for example, it may be the coordinate information (area information) of the in-area and the out-area.

[0067] Modification Example 5: In the above embodiment, various parameters (for the multi-camera object tracking process with path restrictions in the actual operation) are automatically set from the statistical information of the multi-camera object tracking results without path restrictions and the statistical information of the single-camera object tracking results. However, the object tracking parameter setting support system of the present invention is not limited to this. Similar to the above, after automatically setting various parameters from the statistical information of the multi-camera object tracking without path restrictions and the single-camera object tracking results, multi-camera object tracking with restrictions on the movable path is performed using these parameters. Then, from the statistical information of the multi-camera object tracking results with restrictions on the movable path and the statistical information of the single-camera object tracking results, various parameters (for the multi-camera object tracking process with path restrictions in the actual operation) may be reset. Also, various parameters for the multi-camera object tracking process with path restrictions in the actual operation may be automatically set from the statistical information of the multi-camera object tracking results with path restrictions collected in advance and the statistical information of the single-camera object tracking results.

[0068] Modification Example 6: In the above embodiment, for the captured images of each fixed camera 3 during a predetermined period, single camera object tracking is performed, and from the statistical information of these tracking results, in the shooting range 61 of each fixed camera 3, the area where the person to be tracked can pass, or the region of interest (ROI) that is the area where the person to be tracked cannot pass is estimated, and the information of the estimated region of interest is set as one of the parameters used for the multi-camera object tracking. However, the object tracking system of the present invention is not limited to this, and the information of the region of interest estimated from the statistical information of the single camera object tracking results can be used for various applications.

Explanation of Signs

[0069] 1 Analysis server (computer) 3 Fixed camera (camera) 10 Multi-camera object tracking system 37 Parameter setting support program for object tracking 38 Multi-camera object tracking program 40 Parameter setting support system for object tracking 41 Input unit (input means) 43 Mapping unit (mapping means) 44 Single camera object tracking unit (single camera object tracking means) 45 In / Out area estimation unit (In / Out area estimation means) 46 Region of interest estimation unit (region of interest estimation means) 47 Multi-camera object tracking unit without path restriction (multi-camera object tracking means) 48 Edge information estimation unit (movable path estimation means) 49 Entrance / Exit estimation unit (entrance / exit estimation means) 50 Setting unit (setting means) 51 Multi-camera object tracking unit with path restriction (multi-camera object tracking means) 60 2D store map (map of the facility)

Claims

1. An object tracking parameter setting support system that obtains and sets parameters used for multi-camera object tracking, comprising: input means for inputting captured images from a plurality of cameras; multi-camera object tracking means for performing multi-camera object tracking to track a plurality of objects to be tracked across the shooting ranges of the plurality of cameras based on the captured images from the plurality of cameras; movable path estimation means for performing multi-camera object tracking by the multi-camera object tracking means on the captured images for a predetermined period from the plurality of cameras, and estimating a movable path, which is a path between cameras where a person can move, from the statistical information of the tracking results; a setting means for setting the information of the movable path estimated by the movable path estimation means as one of the parameters used for the multi-camera object tracking.

2. single-camera object tracking means for performing single-camera object tracking to track one or more objects to be tracked within the shooting range of one of the plurality of cameras based on the captured image from one of the plurality of cameras; in / out area estimation means for performing single-camera object tracking by the single-camera object tracking means on the captured images for a predetermined period of each of the plurality of cameras, and estimating an in-area, which is an area where a person enters, and an out-area, which is an area where a person exits, in the shooting range of each of the plurality of cameras from the statistical information of these tracking results; and the object tracking parameter setting support system according to claim 1, wherein the setting means sets the information of the in-area and the out-area estimated by the in / out area estimation means as one of the parameters used for the multi-camera object tracking.

3. further comprising entrance / exit estimation means for performing multi-camera object tracking by the multi-camera object tracking means on the captured images for a predetermined period from the plurality of cameras, and estimating an entrance and an exit of the facility where the plurality of cameras are arranged from the statistical information of the tracking results. The object tracking parameter setting support system according to claim 2, wherein the setting means sets the information on the entrance and the exit estimated by the entrance / exit estimation means as one of the parameters used for the multi-camera object tracking.

4. The system further comprises mapping means for associating each point in the captured images of each of the plurality of cameras with each point in a map of the facility where the plurality of cameras are arranged. The object tracking parameter setting support system according to claim 1, wherein the setting means sets the information on the association result by the mapping means as one of the parameters used for the multi-camera object tracking.

5. The movable path estimation means performs multi-camera object tracking by the multi-camera object tracking means on the captured images of the plurality of cameras for a predetermined period without restricting the movable path, which is a path between the cameras where the person can move, and removes, from the candidates for the movable path, the paths between the cameras through which few people have passed as a result of this tracking, thereby estimating the movable path. The object tracking parameter setting support system according to claim 1 is characterized by this.

6. A multi-camera object tracking system that performs multi-camera object tracking on an object captured in the captured images from a plurality of cameras. The multi-camera object tracking system includes multi-camera object tracking means for performing multi-camera object tracking to track a plurality of objects to be tracked across the shooting ranges of the plurality of cameras based on the captured images from the plurality of cameras. A multi-camera object tracking system that performs multi-camera object tracking with restrictions on the movable path on the captured images from the plurality of cameras by the multi-camera object tracking means using the information on the movable path set by the object tracking parameter setting support system according to claim 1.

7. Single-camera object tracking means for performing single-camera object tracking to track one or more objects to be tracked within the shooting range of one camera based on the captured image from one camera. For the captured images of the one camera during a predetermined period, single-camera object tracking is performed by the single-camera object tracking means, and from the statistical information of this tracking result, an object-of-interest region estimating means for estimating a region through which an object to be tracked can pass or a region through which an object to be tracked cannot pass in the imaging range of the one camera is provided. An object tracking system.

8. An object tracking parameter setting support program for obtaining and setting parameters used for multi-camera object tracking, A computer, Input means for inputting captured images from a plurality of cameras, Multi-camera object tracking means for performing multi-camera object tracking to track a plurality of objects to be tracked across the imaging ranges of the plurality of cameras based on the captured images from the plurality of cameras, For the captured images of the plurality of cameras during a predetermined period, multi-camera object tracking is performed by the multi-camera object tracking means, and from the statistical information of this tracking result, a movable path estimating means for estimating a movable path that is a path between cameras through which a person can move is provided. An object tracking parameter setting support program for causing the computer to function as setting means for setting information on the movable path estimated by the movable path estimating means as one of the parameters used for the multi-camera object tracking.

9. A multi-camera object tracking program for performing multi-camera object tracking on an object captured in the captured images from a plurality of cameras, A computer, Function as multi-camera object tracking means for tracking a plurality of objects to be tracked across the imaging ranges of the plurality of cameras based on the captured images from the plurality of cameras, Using the information on the movable path set using the object tracking parameter setting support program according to claim 8, the multi-camera object tracking means performs multi-camera object tracking with restrictions on the movable path for the captured images from the plurality of cameras. A multi-camera object tracking program.

10. An object tracking program for performing object tracking on an object reflected in a captured image from one camera, comprising: a computer, single camera object tracking means for performing single camera object tracking to track one or more objects to be tracked within the shooting range of the one camera based on the captured image from the one camera; interest area estimation means for performing single camera object tracking by the single camera object tracking means on the captured images of the one camera for a predetermined period, and estimating, from the statistical information of the tracking results, an area where an object to be tracked can pass or an area of interest that is an area where an object to be tracked cannot pass in the shooting range of the one camera. An object tracking program comprising.

Images