Mobile object analysis system

Abstract

A mobile object analysis system is provided that can appropriately analyze the movement state of a mobile object at an intersection and in the vicinity thereof while reducing the burden on a user. [Solution] The analysis device 10 of the moving body analysis system 1 acquires a time series of moving body images 2x, acquires movement characteristic information representing the movement characteristics of the moving body, determines whether the new movement characteristic information has a predetermined movement state similarity with the similar characteristic group information, and if the new movement characteristic information has a predetermined movement state similarity with the similar characteristic group information, adds the moving body corresponding to the new movement characteristic information to the similar characteristic group, and updates and stores the similar characteristic group information by reflecting the new movement characteristic information in the similar characteristic group information.

Description

[Technical Field]

[0001] The present invention relates to a mobile object analysis system that analyzes the movement state of a mobile object moving at or near an intersection. [Background technology]

[0002] In recent years, a system for analyzing vehicles passing through an intersection has become known, as described in Patent Document 1. In this system, images of the intersection are continuously captured, and each vehicle in the captured images is recognized using a predetermined image recognition method, thereby creating a bounding box corresponding to each vehicle with an ID. Furthermore, while the image of the intersection is displayed on a display, a pair of passing lines can be set at the entrances and exits of the intersection in the image by a user operating the keyboard / mouse.

[0003] Then, when the center of gravity of the bounding box of any ID passes through the upstream passing line of a pair of passing lines set at the entrance / exit of an intersection, and the midpoint of the bottom of the bounding box passes through the downstream passing line, it is determined that the vehicle corresponding to the bounding box of that ID has passed through the entrance / exit point. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent Publication No. 2021-86467 Summary of the Invention [Problem to be solved by the invention]

[0005] According to the conventional system described above, since a pair of passing lines is set at the entrances and exits of an intersection in an image by a user operating a keyboard / mouse, if the pair of passing lines is not set properly due to human error or the like, it is not possible to properly determine the passage of a moving object such as a vehicle, and as a result, it is not possible to properly analyze the movement state of the moving object. For the same reason, the burden on the user increases as manual operation is required by the user.

[0006] The present invention has been made to solve the above-mentioned problems, and aims to provide a mobile object analysis system that can appropriately analyze the movement status of a mobile object at intersections and their vicinity while reducing the burden on the user. [Means for solving the problem]

[0007] In order to achieve the above object, the moving body analysis system according to claim 1 includes a moving body image acquisition unit that acquires a time series of moving body images, which are images of moving bodies, by a predetermined image recognition process from a traffic environment image that includes a road area including an intersection and moving bodies moving in the road area; a moving body information acquisition unit that acquires moving body information that represents the moving characteristics of the moving body based on the time series of moving body images; a storage unit that stores similar characteristic groups that group one or more moving bodies whose moving body information has a predetermined similarity in moving state, and similar characteristic group information that is data that reflects the moving body information of one or more moving bodies that belong to the similar characteristic group; and a storage unit that stores the newly acquired moving body information, which is new moving body information, each time the moving body information is newly acquired. The system includes a similarity determination unit that determines whether new movement characteristic information has a predetermined movement state similarity with the similar characteristic group information stored in the memory unit, and a management unit that, when the new movement characteristic information has a predetermined movement state similarity with the similar characteristic group information based on the determination result of the similarity determination unit, adds the moving body corresponding to the new movement characteristic information to the similar characteristic group and updates the similar characteristic group information by reflecting the new movement characteristic information in the similar characteristic group information, and is characterized in that the memory unit stores the similar characteristic group to which the moving body corresponding to the new movement characteristic information has been added by the management unit and the updated similar characteristic group information as the similar characteristic group and the similar characteristic group information.

[0008] According to this mobile object analysis system, a time series of mobile object images, which are images of mobile objects, is acquired from a traffic environment image. Based on the time series of mobile object images, mobile object information representing the mobile object's mobility characteristics is acquired. Each time new mobile object information is acquired, it is determined whether the newly acquired new mobile object information has a predetermined similarity in its mobility state with the similar property group information stored in the storage unit. If the determination result indicates that the new mobile object information has a predetermined similarity in its mobility state with the similar property group information, the mobile object corresponding to the new mobile object information is added to the similar property group, and the similar property group information is updated by reflecting the new mobile object information in the similar property group information. As described above, each time new mobile object information is acquired, the mobile object corresponding to the new mobile object information is added to the similar property group and the similar property group information is updated without requiring manual operation by the user. This reduces the burden on the user and allows the mobile object's mobility state to be automatically and appropriately analyzed. In this specification, "image of a mobile object" refers to an image including a mobile object, and "system" refers to a system composed of one or more devices.

[0009] In the present invention, it is preferable that the movement characteristic information acquisition unit has a movement state determination unit that determines whether the movement state of the moving body is a straight-line state or a turning state based on a time series of moving body images, and a final movement characteristic information acquisition unit that, based on the determination result of the movement state determination unit, acquires straight-line movement characteristic information as the movement characteristic information when the movement state of the moving body is a straight-line state, and acquires turning-system movement characteristic information as the movement characteristic information when the movement state of the moving body is a turning state.

[0010] According to this mobile object analysis system, it is determined whether the mobile object is moving in a straight-line state or a turning state based on a time series of images of the mobile object. When the mobile object is moving in a straight-line state, straight-line movement characteristic information is acquired as the movement characteristic information, and when the mobile object is moving in a turning state, turning movement characteristic information is acquired as the movement characteristic information. This makes it possible to create similar characteristic group information and similar characteristic groups for the mobile object while distinguishing between mobile objects moving straight through an intersection and mobile objects turning right or left.

[0011] In the present invention, when the moving state determination unit defines a line segment connecting a predetermined point on the oldest moving body image in the time series of moving body images and a predetermined point on the latest moving body image in the time series of moving body images as a moving line segment, and defines a predetermined point on the moving body image between the predetermined point on the oldest moving body image in the time series of moving body images and the predetermined point on the latest moving body image as a first predetermined point, it is preferable that the moving state determination unit determines whether the moving body is in a straight-line state or a turning state based on information on the figure defined by the moving line segment and the first predetermined point.

[0012] According to this moving body analysis system, it is determined whether the moving body is in a straight-line state or a turning state based on information about a figure defined by a moving line segment and a first predetermined point. Here, the moving line segment is a line segment connecting a predetermined point on the oldest moving body image in the time series of moving body images to a predetermined point on the newest moving body image in the time series of moving body images, and the first predetermined point is a predetermined point on the moving body image between the predetermined point on the oldest moving body image in the time series of moving body images and the predetermined point on the newest moving body image. Therefore, the information about the figure defined by the moving line segment and the first predetermined point appropriately indicates whether the moving body is in a straight-line state or a turning state. Therefore, according to the above method, it is possible to accurately determine whether the moving body is in a straight-line state or a turning state.

[0013] In the present invention, the final movement characteristic information acquisition unit acquires, as straight-line movement characteristic information, a first azimuth angle, which is the angle of a movement line segment, which is a line segment connecting a start point, which is a predetermined point of the oldest moving body image in the time series of moving body images, and an end point, which is a predetermined point of the latest moving body image in the time series of moving body images, with respect to a predetermined reference line, and a center of gravity, which is the midpoint between the start point and the end point; the storage unit stores, as similar characteristic group information, an average center of gravity, which is the average position of the centers of gravity of one or more moving bodies, and a first average azimuth angle, which is the average value of the first azimuth angles; and when the new movement characteristic information is straight-line movement characteristic information, the similarity determination unit determines whether the center of gravity of the new movement characteristic information falls within a predetermined center of gravity range based on the average center of gravity of the similar characteristic group information. When the new movement characteristic information is within the range and the first azimuth angle of the new movement characteristic information is within a first predetermined angle range based on the first average azimuth angle of the similar characteristic group information, the management unit determines that the new movement characteristic information has a predetermined movement state similarity with the similar characteristic group information, and when the new movement characteristic information is straight-line movement characteristic information and the new movement characteristic information has a predetermined movement state similarity with the similar characteristic group information, it is preferable that the management unit adds the moving body corresponding to the new movement characteristic information to the similar characteristic group and calculates, as the similar characteristic group information, the average center of gravity and the first average azimuth angle of the moving bodies belonging to the similar characteristic group to which the moving body corresponding to the new movement characteristic information has been added.

[0014] According to this moving object analysis system, the center of gravity and the first azimuth angle are acquired as the straight-line movement characteristic information, and if the center of gravity of the new movement characteristic information is within a predetermined center of gravity range based on the average center of gravity of the similar characteristic group information and the first azimuth angle of the new movement characteristic information is within a predetermined first azimuth angle range based on the first average azimuth angle of the similar characteristic group information, the new movement characteristic information is determined to have a predetermined movement state similarity with the similar characteristic group information. Here, the first azimuth angle is the angle with respect to a predetermined reference line of a movement line segment connecting a start point, which is a predetermined point of the oldest moving object image in the time series of moving object images, and an end point, which is a predetermined point of the latest moving object image in the time series of moving object images, and the center of gravity is the midpoint between the start point and the end point, these first azimuth angle and the center of gravity can be considered to be information that appropriately represents the movement characteristics of a moving object in a straight-line state.

[0015] Therefore, when the center of gravity of the new movement characteristic information is within a predetermined center of gravity range based on the average center of gravity of the similar characteristic group information, and the first azimuth angle of the new movement characteristic information is within a first predetermined angle range based on the first average azimuth angle of the similar characteristic group information, the moving body corresponding to the new movement characteristic information can be considered to be in a straight-line movement state similar to that of a moving body belonging to the similar characteristic group, and it can be appropriately determined that the new movement characteristic information has a predetermined movement state similarity with the similar characteristic group information.

[0016] Furthermore, when the new movement characteristic information has a predetermined movement state similarity with the similar characteristic group information, the moving body corresponding to the new movement characteristic information is added to the similar characteristic group, and the average center of gravity and first average azimuth angle of the moving bodies belonging to the similar characteristic group to which the moving body corresponding to the new movement characteristic information has been added are calculated as the similar characteristic group information, thereby allowing the similar characteristic group information to be calculated as information that reflects the new movement characteristic information in the original similar characteristic group information.

[0017] In the present invention, the final movement characteristic information acquisition unit acquires, as turning system movement characteristic information, a start point, which is a predetermined point of the oldest moving body image in the time series of moving body images, a second azimuth angle, which is the angle of the azimuth line, which is a line connecting the start point and a predetermined point of the moving body image next to the oldest moving body image, with respect to a predetermined reference line, an end point, which is a predetermined point of the latest moving body image in the time series of moving body images, and a separation point, which is a point among the predetermined points in the time series of moving body images that is the longest distance from the movement line, which is a line connecting the start point and the end point; the memory unit stores, as similar characteristic group information, an average start point, which is the average position of the start points of one or more moving bodies, a second average azimuth angle, which is the average value of the second azimuth angle, an average separation point, which is the average position of the separation points, and an average end point, which is the average position of the end points; and the similarity determination unit determines, when the new movement characteristic information is turning system movement characteristic information, that the start point in the new movement characteristic information is within a predetermined start point range based on the average start point of the similar characteristic group information; If all of the following are true: the second azimuth angle of the new movement characteristic information is within a second predetermined angle range based on the second average azimuth angle of the similar characteristic group information; the departure point of the new movement characteristic information is within a predetermined departure point range based on the average departure point of the similar characteristic group information; and the end point of the new movement characteristic information is within a predetermined end point range based on the average end point of the similar characteristic group information, the new movement characteristic information is determined to have a predetermined movement state similarity with the similar characteristic group information, and when the new movement characteristic information is turning-type movement characteristic information and the new movement characteristic information has a predetermined movement state similarity with the similar characteristic group information, it is preferable that the management unit adds the moving body corresponding to the new movement characteristic information to the similar characteristic group and calculates, as similar characteristic group information, the average starting point, second average azimuth angle, average departure point, and average end point of the moving bodies belonging to the similar characteristic group to which the moving body corresponding to the new movement characteristic information has been added.

[0018] According to this mobile object analysis system, the starting point, second azimuth angle, end point, and departure point are acquired as turning-type movement characteristic information, and when the new movement characteristic information is turning-type movement characteristic information, if all of the following are true: the starting point in the new movement characteristic information is within a predetermined starting point range based on the average starting point of the similar characteristic group information, the second azimuth angle of the new movement characteristic information is within a second predetermined angle range based on the second average azimuth angle of the similar characteristic group information, the departure point of the new movement characteristic information is within a predetermined departure point range based on the average departure point of the similar characteristic group information, and the end point of the new movement characteristic information is within a predetermined end point range based on the average end point of the similar characteristic group information, then the new movement characteristic information is determined to have a predetermined movement state similarity with the similar characteristic group information.

[0019] Here, the starting point is a predetermined point of the oldest moving body image in the time series of moving body images, the second azimuth angle is the angle of the azimuth line segment, which is a line segment connecting the starting point and a predetermined point of the moving body image next to the oldest moving body image, relative to a predetermined reference line, the ending point is a predetermined point of the latest moving body image in the time series of moving body images, and the separation point is a point among the predetermined points in the time series of moving body images that is the longest distance from the moving line segment, which is a line segment connecting the starting point and the ending point.Therefore, these parameters can be considered as information that appropriately represents the movement characteristics of a moving body in a turning state.

[0020] Therefore, when the starting point in the new movement characteristic information is within a predetermined starting point range based on the average starting point of the similar characteristic group information, the second azimuth angle of the new movement characteristic information is within a second predetermined angle range based on the second average azimuth angle of the similar characteristic group information, the departure point of the new movement characteristic information is within a predetermined departure point range based on the average departure point of the similar characteristic group information, and the end point of the new movement characteristic information is within a predetermined end point range based on the average end point of the similar characteristic group information, the moving body corresponding to the new movement characteristic information can be considered to be in a turning movement state similar to that of a moving body belonging to the similar characteristic group, and it can be appropriately determined that the new movement characteristic information has a predetermined movement state similarity with the similar characteristic group information.

[0021] Furthermore, when the new movement characteristic information is turning-type movement characteristic information and the new movement characteristic information has a predetermined movement state similarity with the similar characteristic group information, the moving body corresponding to the new movement characteristic information is added to the similar characteristic group, and the average starting point, second average azimuth angle, average separation point, and average ending point of the moving bodies belonging to the similar characteristic group to which the moving body corresponding to the new movement characteristic information has been added are calculated as the similar characteristic group information, thereby allowing the similar characteristic group information to be calculated as information that reflects the new movement characteristic information in the original similar characteristic group information.

[0022] In the present invention, when the new movement characteristic information does not have similarity with the similar characteristic group information, the management unit sets the moving body corresponding to the new movement characteristic information as a moving body belonging to a new similar characteristic group and sets the new movement characteristic information as new similar characteristic group information, and the memory unit further stores new similar characteristic groups and new similar characteristic group information in addition to the similar characteristic groups and similar characteristic group information, and it is preferable that the similarity determination unit determines whether the new movement characteristic information has a predetermined movement state similarity with the similar characteristic group information and new similar characteristic group information stored in the memory unit each time new movement characteristic information is acquired.

[0023] According to this mobile object analysis system, when the new movement characteristic information of a mobile object does not have similarity with the similar characteristic group information, the mobile object is set as a mobile object belonging to a new similar characteristic group, and the new movement characteristic information is set as new similar characteristic group information, thereby making it possible to appropriately group and set information of mobile objects with different movement characteristics, thereby making it possible to more appropriately analyze the movement states of mobile objects moving in multiple directions within an intersection.

[0024] In the present invention, it is preferable to further include a counting unit that counts the number of one or more moving objects belonging to a similar characteristic group, and an output unit that outputs counting information representing the number of one or more moving objects counted by the counting unit.

[0025] According to this mobile object analysis system, the number of one or more mobile objects belonging to a similar characteristic group is counted and coefficient information representing this is output, thereby allowing the user to recognize the number of one or more mobile objects belonging to a similar characteristic group.

[0026] In the present invention, the storage unit further stores, as similar characteristic group information, a characteristic vector extending from the average center of gravity along a first average azimuth angle, and when a plurality of similar characteristic group information is stored in the storage unit, a driving lane generation unit generates, as a driving lane, a straight line extending at the first average azimuth angle while passing through the average center of gravity in each of the plurality of similar characteristic group information; an intersection generation unit generates an intersection of two driving lanes in the plurality of driving lanes when the intersection angle between two driving lanes is within a range of 90°±γ (γ is a positive value); a first point setting unit sets, when two intersections are generated on one driving lane, one of the two intersections as an entrance side intersection on the side where the mobile body enters and the other of the two intersections as an exit side intersection on the side where the mobile body leaves, based on the similar characteristic group information corresponding to one driving lane; It is preferable that the vehicle control system further comprises a second point setting unit that sets a point a predetermined value away from the intersection on the opposite side of the intersection as a first entry point, sets a point a predetermined value away from the entry-side intersection on a driving lane that intersects with one of the driving lanes at the entry-side intersection in the direction of the characteristic vector as a first departure point, sets a point a predetermined value away from the departure-side intersection on a driving lane that intersects with one of the driving lanes at the departure-side intersection in the direction opposite to the characteristic vector as a second entry point, and sets a point a predetermined value away from the departure-side intersection on one of the driving lanes on the opposite side of the entry-side intersection as a second departure point, and a curved road generation unit that generates a first curved road connecting the first entry point and the first departure point as a parametric curved road that has a maximum curvature at a portion closest to the entry-side intersection, and generates a second curved road connecting the second entry point and the second departure point as a parametric curved road that has a maximum curvature at a portion closest to the departure-side intersection.

[0027] According to this mobile object analysis system, straight lines passing through the average center of gravity in each of the four similar characteristic group information and extending in the direction of the first average azimuth angle are generated as four driving lanes, and if the intersection angle between two driving lanes is within the range of 90°±γ (γ is a positive value), an intersection between the two driving lanes is generated, and if two intersections are generated on one driving lane, one of the two intersections is set as the entry side intersection on the side where the mobile object enters, and the other of the two intersections is set as the exit side intersection on the side where the mobile object leaves, based on the similar characteristic group information corresponding to one driving lane. Furthermore, a point on one of the driving lanes that is spaced a predetermined value from the entry-side intersection on the opposite side of the exit-side intersection is set as a first entry point, a point on the driving lane that intersects with one of the driving lanes at the entry-side intersection on the direction of the characteristic vector on the entry-side intersection on the driving lane that intersects with one of the driving lanes at the exit-side intersection on the direction opposite to the characteristic vector on the entry-side intersection on the driving lane that intersects with one of the driving lanes at the exit-side intersection on the direction opposite to the entry-side intersection on the driving lane is set as a second entry point, and a point on the driving lane that is spaced a predetermined value from the exit-side intersection on the opposite side of the entry-side intersection on the driving lane on the direction opposite to the entry-side intersection on the entry-side intersection on the driving lane is set as a second departure point. A first curved path connecting the first entry point and the first departure point is generated as a parametric curved path with a maximum curvature at a portion closest to the entry-side intersection, and a second curved path connecting the second entry point and the second departure point is generated as a parametric curved path with a maximum curvature at a portion closest to the exit-side intersection on the driving lane. As described above, by generating the driving lane and the first and second exit curved paths, it is possible to generate the driving path and left-turn path of a moving object at an intersection. As a result, if security areas are set along the driving path of the moving object, it is possible to determine the direction in which the moving object will enter those security areas. Furthermore, if a crosswalk is present at the intersection, it is possible to determine the possibility that the moving object will enter the crosswalk after turning left and the degree of proximity of the moving object to the crosswalk. [Brief explanation of the drawings]

[0028] [Figure 1] 1 is a diagram schematically illustrating a configuration of a moving object analysis system according to an embodiment of the present invention. [Figure 2]FIG. 1 is a diagram showing an image of an intersection and its vicinity captured by a camera. [Figure 3] FIG. 2 is a block diagram showing the functional configuration of the analysis device. [Figure 4] FIG. 10 is a diagram illustrating an example of a moving object image. [Figure 5] FIG. 10 is an explanatory diagram of a method for determining the moving state of a moving object. [Figure 6] FIG. 10 is a diagram showing an example of an acquired result of straight-line movement characteristic information. [Figure 7] FIG. 10 is a diagram illustrating an example of an acquired result of turning movement characteristic information. [Figure 8] 10 is a diagram for explaining whether or not new movement characteristic information has similarity to straight-line similar characteristic group information. FIG. [Figure 9] 10 is a diagram for explaining whether or not new movement characteristic information has similarity to turning-related similar characteristic group information. FIG. [Figure 10] FIG. 10 is a diagram showing an example of a result of setting a similar characteristic group of a linear system. [Figure 11] FIG. 10 is a diagram showing an example of a result of generating a travel route. [Figure 12] FIG. 10 is a diagram showing an example of a straight-line counting result image. [Figure 13] FIG. 10 is a diagram showing another example of the straight-line counting result image. [Figure 14] FIG. 10 is a diagram illustrating an example of a travel route image. DETAILED DESCRIPTION OF THE INVENTION

[0029] A mobile object analysis system according to one embodiment of the present invention will be described below with reference to the drawings. The mobile object analysis system 1 shown in Fig. 1 of this embodiment is for analyzing a mobile object 8 (see Fig. 2) such as a vehicle passing through an intersection 5, and includes a camera 3 and an analysis device 10.

[0030] The analysis device 10 is configured as a computer equipped with a CPU, storage, memory, and an I / O interface, and is capable of communicating with the camera 3. In the analysis device 10, an analysis process of the moving object is performed based on the image signal from the camera 3, as will be described later.

[0031] Furthermore, an external terminal 20 is communicatively connected to the analysis device 10, and the external terminal 20 is configured as, for example, a personal computer. As will be described later, the external terminal 20 displays data received from the analysis device 10 on a display (not shown).

[0032] In this embodiment, the intersection 5 is configured in the shape of a crossroad where a first road 5a and a second road 5b intersect in a cross shape. Pedestrian crossings 5c, 5c are provided in the vicinity of the intersection 5 of the first road 5a, and pedestrian crossings 5d, 5d are provided in the vicinity of the intersection 5 of the second road 5b.

[0033] Camera 3 is placed diagonally above intersection 5 and has a communication function. Camera 3 captures image 2 (hereinafter referred to as "traffic environment image 2") including intersection 5, part of first road 5a, and part of second road 5b at an angle as shown in Fig. 2, and transmits the image signal to analysis device 10. In this case, traffic environment image 2 is captured as a video.

[0034] Next, the functional configuration of the analysis device 10 will be described with reference to Fig. 3. As shown in Fig. 3, the analysis device 10 has, as its functions, a moving object image acquisition unit 11, a movement characteristic information acquisition unit 12, a similarity determination unit 13, a management unit 14, a storage unit 15, a counting unit 16, a movement route generation unit 17, and an output unit 18.

[0035] First, in the moving object image acquisition unit 11, when the image signal from the camera 3 is received by the analysis device 10, a moving object image 2x as shown in Fig. 4 is acquired frame by frame as a still image from the traffic environment image 2. That is, a time series of the moving object images 2x is acquired.

[0036] 4, in the moving body image 2x, a bounding box 30 is acquired by a predetermined image recognition process so as to surround a moving body 8 such as a four-wheeled vehicle in each frame of the traffic environment image 2, and the midpoint P of the bottom side of the bounding box 30 is acquired. In this case, an object recognition process applying a machine learning model (for example, OpenVino: a registered trademark) is used as the predetermined image recognition process.

[0037] Next, the movement characteristic information acquisition unit 12 will be described. As will be described below, this movement characteristic information acquisition unit 12 acquires movement characteristic information of the moving body 8 from n (n is plural) moving body images 2x, and includes a movement state determination unit 12a and a final movement characteristic information acquisition unit 12b. In the following description, movement characteristic information newly acquired from the moving body image 2x by the movement characteristic information acquisition unit 12 will be referred to as "new movement characteristic information."

[0038] As will be described below, this moving state determination unit 12a determines whether the moving state of the moving object 8 is a straight-line state or a turning state. In this moving state determination unit 12a, first, midpoints P of n bounding boxes 30 of the moving object 8 are acquired from n moving object images 2x, and these n midpoints Pi (i=1 to n) are arranged as shown in Fig. 5. Note that Fig. 5 shows an example where n=5.

[0039] In the figure, the starting point P1 represents the midpoint P in the oldest frame, and the ending point Pn represents the midpoint P in the latest frame. Next, a triangle is created with a moving line segment Lm, which is a line segment connecting the starting point P1 and the ending point Pn, as its base, and points P2 to P4 other than the starting point P1 and the ending point Pn as its vertices. Furthermore, of the vertices P2 to P4 in each triangle, the vertex that is the greatest distance from the moving line segment Lm is defined as a separation point Pm (first predetermined point), and the distance Dm between this separation point Pm and the moving line segment Lm is calculated. In the example shown in FIG. 5, point P3 is the separation point Pm.

[0040] When the distance Dm is greater than the predetermined value Dref, it is determined that the moving state of the moving object 8 is in a turning state, and when the distance Dm is equal to or less than the predetermined value Dref, it is determined that the moving state of the moving object 8 is in a straight-line moving state. By the above method, the moving state determination unit 12a determines whether the moving state of the moving object 8 is in a straight-line moving state or a turning state. Note that Fig. 5 shows an example where the moving state of the moving object 8 is in a turning state.

[0041] On the other hand, as described below, the final movement characteristic information acquisition unit 12b acquires straight-line movement characteristic information or turning-line movement characteristic information as new movement characteristic information based on the determination result of the movement state of the moving body 8 by the movement state determination unit 12a. In this case, when the movement state of the moving body 8 is a straight-line state, straight-line movement characteristic information is acquired, and when the movement state of the moving body 8 is a turning state, turning-line movement characteristic information is acquired.

[0042] First, the straight-line movement characteristic information will be described with reference to Fig. 6. In this case, as the straight-line movement characteristic information, the first azimuth angle θ1 and the center of gravity G are acquired as described below. First, as shown in Fig. 6, n midpoints Pi (i = 1 to n) acquired by the above-mentioned method are arranged to create a movement vector Vm (movement line segment) extending from a start point P1 to an end point Pn. Fig. 6 shows an example where n = 5.

[0043] Next, when a line extending in the left-right direction of the moving object image 2x is taken as a predetermined reference line Lb, a first azimuth angle θ1 is acquired as the angle between the movement vector Vm and the predetermined reference line Lb. In this case, the first azimuth angle θ1 is acquired as a positive value when it is a clockwise angle with respect to the predetermined reference line Lb, and as a negative value when it is a counterclockwise angle with respect to the predetermined reference line Lb. Therefore, in the example of FIG. 6, the first azimuth angle θ1 is acquired as a negative value. This also applies to various angles described below. In addition, the center of gravity G is acquired as the midpoint between the start point P1 and the end point Pn. Note that the same figure shows an example where G=P3.

[0044] Next, turning system movement characteristic information will be described with reference to Fig. 7. In this case, as the turning system movement characteristic information, a start point P1, an end point Pn, a second azimuth angle θ2, and a separation point Pm are acquired, as described below. First, as shown in Fig. 7, n midpoints Pi (i = 1 to n) acquired by the above-mentioned method are arranged to create a movement vector Vm extending from the start point P1 to the end point Pn. Fig. 7 shows an example where n = 5.

[0045] Next, when the line segment connecting the start point P1 and the next midpoint P2 of the start point P1 is defined as the azimuth line segment Ld, the second azimuth angle θ2 is obtained as the angle between the azimuth line segment Ld and the predetermined reference line Lb. Furthermore, the separation point Pm is obtained as the midpoint Pi at which the distance from the movement vector Vm is maximum. Figure 7 shows an example where the separation point Pm=P3.

[0046] Next, the similarity determination unit 13 will be described. The similarity determination unit 13 determines whether or not the new movement characteristic information acquired by the movement characteristic information acquisition unit 12 has a predetermined similarity in movement state with respect to similar characteristic group information in the similar characteristic group stored in the storage unit 15. In this case, as will be described later, the storage unit 15 stores similar characteristic group information in the straight-line similar characteristic group (hereinafter referred to as "straight-line similar characteristic group information") and similar characteristic group information in the turning similar characteristic group (hereinafter referred to as "turning similar characteristic group information").

[0047] First, referring to Fig. 8, an example will be described in which the moving object 8 is in a straight-line movement state and the new movement characteristic information is straight-line movement characteristic information. Here, when the new movement characteristic information is straight-line movement characteristic information, the average center of gravity Gave and the first average azimuth angle θ1ave are used as straight-line similar characteristic group information.

[0048] As will be described later, the average center of gravity Gave is the average position of the center of gravity G in the straight-line movement characteristic information of moving bodies 8 that belong to one straight-line similar characteristic group. Also, the first average azimuth angle θ1ave is the average value of the first azimuth angle θ1 in the straight-line movement characteristic information of moving bodies 8 that belong to one straight-line similar characteristic group.

[0049] When all of the following conditions (f1) to (f2) are met, it is determined that the new movement characteristic information has a predetermined movement state similarity to the similar characteristic group information, and otherwise it is determined that the new movement characteristic information does not have a predetermined movement state similarity to the similar characteristic group information.

[0050] (f1) The center of gravity G in the new movement characteristic information is located within a predetermined center of gravity range Ag (the range indicated by the two-dot chain line in FIG. 8). This predetermined center of gravity range Ag is defined as a circular range centered on the average center of gravity Gave. (f2) θ1ave-α≦θ1≦θ1ave+α (α is a predetermined positive value) is satisfied. That is, the first azimuth angle θ1 in the new movement characteristic information is within a predetermined first range with the value θ1ave-α as the lower limit and the value θ1ave+α as the upper limit.

[0051] Next, referring to Fig. 9, an example will be described in which the moving object 8 is in a turning state and the new movement characteristic information is turning-system movement characteristic information. When the new movement characteristic information is turning-system movement characteristic information, the average start point P1ave, the second average azimuth angle θ2ave, the plane departure point Pmave, and the average end point Pnave are used as turning-system similar characteristic group information. As will be described later, this average start point P1ave is the average position of the start point P1 in the turning-system movement characteristic information of the moving object 8 belonging to one turning-system similar characteristic group.

[0052] Furthermore, the second average azimuth angle θ2ave is the average value of the second azimuth angle θ2 in the turning system movement characteristic information of the moving bodies 8 that belong to one turning system similar characteristic group, as will be described later. Furthermore, the plane separation point Pmave is the average position of the separation point Pm in the turning system movement characteristic information of the moving bodies 8 that belong to one turning system similar characteristic group, as will be described later.

[0053] On the other hand, the average end point Pnave is the average position of the end points Pn in the turning-type movement characteristic information of the moving objects 8 that belong to one turning-type similar characteristic group, as will be described later.

[0054] Then, when all of the following conditions (f3) to (f6) are met, it is determined that the new movement characteristic information has a predetermined movement state similarity to the similar characteristic group information, and otherwise it is determined that the new movement characteristic information does not have a predetermined movement state similarity to the similar characteristic group information.

[0055] (f3) The starting point P1 in the new movement characteristic information is located within a predetermined starting point range Ap1 (the range indicated by the two-dot chain line in FIG. 9). This predetermined starting point range Ap1 is defined as a circular range centered on the average starting point P1ave. (f4) θ2ave-β≦θ2≦θ2ave+β (β is a predetermined positive value) is satisfied. That is, the second azimuth angle θ2 in the new movement characteristic information is within a predetermined second angle range with the value θ2ave-β as the lower limit and the value θ2ave+β as the upper limit.

[0056] (f5) The separation point Pm in the new movement characteristic information is located within a predetermined separation point range Apm (the range indicated by the two-dot chain line in FIG. 9). This predetermined separation point range Apm is defined as a circular range centered on the average separation point Pmave. (f6) The end point Pn in the new movement characteristic information is located within a predetermined end point range Apn (the range indicated by the two-dot chain line in FIG. 9). This predetermined end point range Apn is defined as a circular range centered on the average end point Pnave.

[0057] Next, the management unit 14 will be described. In this management unit 14, as described below, the similar characteristic group information is updated / set based on the determination result of the above-mentioned similarity determination unit 13. First, in the case where the new movement characteristic information is straight-line movement characteristic information, when it is determined that the new movement characteristic information has a predetermined similarity in movement state with respect to the straight-line similar characteristic group information, the moving object corresponding to the new movement characteristic information is added to the straight-line similar characteristic group, and, as described below, the new movement characteristic information is reflected in the similar characteristic group information, thereby updating (calculating) the average center of gravity Gave and the first average azimuth angle θ1ave, which are the straight-line similar characteristic group information.

[0058] In other words, the data of the center of gravity G of the new movement characteristic information is added to the data of the center of gravity G of the movement characteristic information of all moving bodies belonging to the similar characteristic group, thereby obtaining the data of the center of gravity G of the entire group, and the average center of gravity Gave is calculated as the position of the simple average of these data.

[0059] In addition, the first azimuth angle θ1 of the new movement characteristic information is added to the data of the first azimuth angle θ1 of the movement characteristic information of all moving bodies belonging to the similar characteristic group, thereby obtaining data of the first azimuth angle θ1 of the entire group, and the first average azimuth angle θ1ave is calculated as the simple average value of this data.

[0060] On the other hand, when the new movement characteristic information is straight-line movement characteristic information and it is determined that the new movement characteristic information does not have a predetermined movement state similarity with the straight-line similar characteristic group information, the moving body corresponding to the new movement characteristic information is set as a moving body belonging to a new similar characteristic group, and the new movement characteristic information is set as new similar characteristic group information.

[0061] Through the above processing, the management unit 14 sets, for example, linear similar property groups as shown in Fig. 10. Fig. 10 shows an example in which three linear similar property groups Grx to Grz are set, and three moving bodies 8x1 to 8x3 belong to group Grx. In the case of this similar property group Grx, its average center of gravity Gavex is set to the average position of the centers of gravity Gx1 to Gx3 of the three moving bodies 8x1 to 8x3, and the predetermined center-of-gravity range Agx is set to a circular range centered on the average center of gravity Gavex.

[0062] Furthermore, although not shown, the first average azimuth angle θ1avex is calculated as the average value of the first azimuth angles θ1x1 to θ1x3 of the three moving bodies 8x1 to 8x3. If the vector extending from the average center of gravity Gavex to the right in Fig. 10 along a straight line that forms the first average azimuth angle θ1avex with respect to the predetermined reference line Lb is defined as a characteristic vector Vpx, then the above group Grx can be considered as a group of moving bodies with characteristics that can be expressed by the characteristic vector Vpx.

[0063] On the other hand, in the case of group Gry, one moving object 8y belongs to the group, and the center of gravity Gy of the moving object 8y is set to the average center of gravity Gavey, and the predetermined center-of-gravity range Agy is set to a circular range centered on the average center of gravity Gavey. Furthermore, although not shown, a first average azimuth angle θ1avey is calculated as the first azimuth angle θ1y of the moving object 8y. Here, if a vector extending from the average center of gravity Gavey to the right in FIG. 10 along a straight line that forms the first average azimuth angle θ1avey with respect to the predetermined reference line Lb is defined as a characteristic vector Vpy, the above group Gry can be considered as a group of moving objects with characteristics that can be expressed by the characteristic vector Vpy.

[0064] For example, if this moving body 8y is a moving body corresponding to new movement characteristic information, then although θ1avex-α≦θ1y≦θ1avex+α holds at the first azimuth angle θ1y of the moving body 8y, the center of gravity Gy of the moving body 8y is not located within the specified center of gravity range Agx, and therefore the moving body 8y will be set as a moving body belonging to the new similar characteristic group Gry.

[0065] In the case of the similar property group Grz, the average center of gravity Gavez is set to the average position of the centers of gravity Gz1 to Gz3 of the three moving bodies 8z1 to 8z3, and the predetermined center of gravity range Agz is set to a circular range centered on the average center of gravity Gavez.

[0066] Furthermore, although not shown, the first average azimuth angle θ1avez is calculated as the average value of the first azimuth angles θ1z1 to θ1z3 of the three moving bodies 8z1 to 8z3. If a vector extending from the average center of gravity Gavez to the left in the figure along a straight line that forms the first average azimuth angle θ1avez with respect to the predetermined reference line Lb is defined as a characteristic vector Vpz, then the above group Grz can be considered as a group of moving bodies with characteristics that can be expressed by the characteristic vector Vpz.

[0067] Furthermore, when the new movement characteristic information is turning system movement characteristic information and it is determined that the new movement characteristic information has a predetermined movement state similarity with the turning system similar characteristic group information, the moving body corresponding to the new movement characteristic information is added to the turning system similar characteristic group, and as described below, the new movement characteristic information is reflected in the similar characteristic group information, and the turning system similar characteristic group information, namely the average starting point P1ave, the second average azimuth angle θ2ave, the plane separation point Pmave, and the average ending point Pnave, are updated (calculated).

[0068] That is, although not shown in the figure, the data for the starting point P1 of the new movement characteristic information is added to the data for the starting point P1 of the movement characteristic information of all moving bodies belonging to the similar characteristic group, thereby obtaining data for the starting point P1 of the entire group, and the average starting point P1ave is calculated as the position of the simple average of these data.

[0069] Also, although not shown, the data for the second azimuth angle θ2 of the new movement characteristic information is added to the data for the second azimuth angle θ2 of the movement characteristic information of all moving bodies belonging to the similar characteristic group, thereby obtaining data for the second azimuth angle θ2 of the entire group, and the second average azimuth angle θ2ave is calculated as the simple average value of these data.

[0070] Furthermore, although not shown in the figure, the data on the separation points Pm of the new movement characteristic information is added to the data on the separation points Pm of the movement characteristic information of all moving bodies belonging to the similar characteristic group, thereby obtaining data on the separation points Pm of the entire group, and the average separation point Pmave is calculated as the position of the simple average of these data.

[0071] Also, although not shown in the figure, the data for the end point Pn of the new movement characteristic information is added to the data for the end point Pn of the movement characteristic information of all moving bodies belonging to the similar characteristic group, thereby obtaining data for the end point Pn of the entire group, and the average end point Pnave is calculated as the position of the simple average of these data.

[0072] Similarly, when the new movement characteristic information is turning-type movement characteristic information and it is determined that the new movement characteristic information does not have a predetermined movement state similarity with the turning-type similar characteristic group information, the moving body corresponding to the new movement characteristic information is set as a moving body belonging to a new similar characteristic group, and the new movement characteristic information is set as new similar characteristic group information.

[0073] Then, the storage unit 15 stores the similar characteristic group and similar characteristic group information calculated or set as described above by the management unit 14. When the similar characteristic group is a linear similar characteristic group, the direction of the characteristic vector is also stored as the similar characteristic group information in addition to the average center of gravity Gave and the first average azimuth angle θ1ave.

[0074] Furthermore, the counting unit 16 counts the number of moving objects that belong to the straight-line similar characteristic group and the turning similar characteristic group stored in the storage unit 15.

[0075] Next, we will explain the travel route generation unit 17. The travel route generation unit 17 generates a travel route Rm (a route indicated by a dashed line in FIG. 11) as described below, and includes, as its functions, a travel lane generation unit 17a, an intersection generation unit 17b, a first point setting unit 17c, a second point setting unit 17d, and a curved road generation unit 17e, as shown in FIG.

[0076] In the following description, an example will be given in which four straight-line similar characteristic group information items are stored in the storage unit 15. In this case, four groups Gr1 to Gr4 are stored in the storage unit 15 as straight-line similar characteristic groups, and four average centers of gravity Gave1 to Gave4, four first average azimuth angles θ1ave1 to θ1ave4 (none of which are shown), and four characteristic vectors Vp1 to Vp4 are stored as similar characteristic group information items for these groups Gr1 to Gr4.

[0077] First, the driving lane generation unit 17a generates four driving lanes L1 to L4 based on the information on the four straight-line similar characteristic groups, as shown in Fig. 11. That is, the driving lane L1 is generated as a straight line that passes through the average center of gravity Gave1 of the group Gr1 and extends along the characteristic vector Vp1 (i.e., a straight line that extends in the direction of the first average azimuth angle θ1ave1), and the driving lane L2 is generated as a straight line that passes through the average center of gravity Gave2 of the group Gr2 and extends along the characteristic vector Vp2 (i.e., a straight line that extends in the direction of the first average azimuth angle θ1ave2).

[0078] Furthermore, the driving lane L3 is generated as a straight line passing through the average center of gravity Gave3 of group Gr3 and extending along the characteristic vector Vp3 (i.e., a straight line extending in the direction of the first average azimuth angle θ1ave3), and the driving lane L4 is generated as a straight line passing through the average center of gravity Gave4 of group Gr4 and extending along the characteristic vector Vp4 (i.e., a straight line extending in the direction of the first average azimuth angle θ1ave4).

[0079] Next, the intersection point generation unit 17b generates an intersection point between two of the four driving lanes L1 to L4 if the intersection angle between the two driving lanes is within the range of 90°±γ (γ is a positive value). In this case, the determination of whether the intersection angle between the two driving lanes is 90°±γ is made based on whether the dot product of the two characteristic vectors is within a predetermined range.

[0080] In this embodiment, the above conditions are met, and therefore four intersections P13, P14, P23, and P24 are generated. That is, intersections P13 and P14 are generated as intersections between driving lane L1 and driving lanes L3 and L4, and intersections P23 and P24 are generated as intersections between driving lane L2 and driving lanes L3 and L4.

[0081] Furthermore, the first point setting unit 17c sets the intersection P13 as the entrance-side intersection on the side where the moving object enters, and sets the intersection P14 as the departure-side intersection on the side where the moving object leaves, based on the orientations of the intersections P13 and P14 on the driving lane L1 and the characteristic vector Vp1. Furthermore, based on the orientations of the intersections P23 and P24 on the driving lane L2 and the characteristic vector Vp2, the intersection P24 is set as the entrance-side intersection on the side where the moving object enters, and the intersection P23 is set as the departure-side intersection on the side where the moving object leaves.

[0082] Furthermore, the second point setting unit 17d generates a circle C13 of a predetermined diameter centered on the entry intersection point P13, and among the points where this circle C13 intersects with the driving lane L1, the point away from the entry intersection point P13 toward the average center of gravity Gave1 is set as the first entry point Pin13, and among the points where the circle C13 intersects with the driving lane L3, the point away from the entry intersection point P13 on the opposite side of the intersection point P23 is set as the first departure point Pout13.

[0083] In addition, a circle C14 of the same diameter as circle C13 is generated with the exit-side intersection point P14 as its center, and the point at which this circle C14 intersects with the driving lane L4, which is away from the exit-side intersection point P14 toward the average center of gravity Gave4, is set as the second entry point Pin14, and the point at which circle C14 intersects with the driving lane L1, which is away from the exit-side intersection point P14 on the opposite side of the entry-side intersection point P13, is set as the second departure point Pout14.

[0084] Furthermore, using the same method as above, a first approach point Pin24 and a first departure point Pout24 are set corresponding to the approach side intersection P24, and a second approach point Pin23 and a second departure point Pout23 are set corresponding to the departure side intersection P23.

[0085] Then, the curved path generating unit 17e generates four curved paths Lc13, Lc14, Lc23, and Lc24. The first curved path Lc13 is a curved path connecting the first entry point Pin13 and the first departure point Pout13, and is generated as a parametric curve (for example, a Bezier curve) whose curvature is maximized at the portion closest to the entry-side intersection P13.

[0086] The second curved path Lc14 is a curved path connecting the second entry point Pin14 and the second departure point Pout14, and is generated as a parametric curve (for example, a Bezier curve) whose curvature is maximized at the location closest to the departure-side intersection point P14.

[0087] Furthermore, the second curved path Lc23 is a curved path connecting the second entry point Pin23 and the second departure point Pout23, and is generated as a parametric curve (for example, a Bezier curve) whose curvature is maximized at the location closest to the departure-side intersection point P23.

[0088] On the other hand, the first curved path Lc24 is a curved path connecting the first entry point Pin24 and the first departure point Pout24, and is generated as a parametric curve (for example, a Bezier curve) whose curvature is maximized at the portion closest to the entry-side intersection P24.

[0089] The travel route Rm is configured as a combination of the above four travel lanes L1 to L4 and four curved roads Lc13, Lc14, Lc23, and Lc24.

[0090] Next, we will explain the output unit 18. For example, when the output unit 18 receives a signal requesting image data from the external terminal 20, it creates a counting result image signal or a movement route image signal, and transmits either of these image signals to the external terminal 20.

[0091] This counting result image signal is a signal including a straight-line counting result image and a turning-line counting result image that represent the counting results in the aforementioned counting unit 16, and the straight-line counting result image is an image that represents the counting results of moving objects in the straight-line similar characteristic group. Also, the turning-line counting result image is an image that represents the counting results of moving objects in the turning-line similar characteristic group. Also, the movement route image signal is a signal including a movement route image that represents the movement route Rm created by the movement route generation unit 17.

[0092] When the external terminal 20 receives the counting result image signal, the straight-line counting result image and the turning-line counting result image included in the counting result image signal are displayed on a display (not shown) of the external terminal 20. In the following, the straight-line counting result image 40 will be described with reference to FIG. 12.

[0093] 12, in the straight-line counting result image 40, four circular counting display frames 41 are displayed at the positions of the four average centers of gravity, and the counting values ​​(values ​​3, 6, 20, and 21 in FIG. 12) of the moving objects belonging to each straight-line similar characteristic group are displayed in each counting display frame 41. This allows the user of the external terminal 20 to grasp the traffic volume of moving objects moving straight in each driving lane.

[0094] In this case, as shown in Fig. 13, the straight-line counting result image 40 may be configured to display four rectangular display frames 42 at the positions of the four average centers of gravity. In addition to the count values ​​of the moving objects belonging to each straight-line similar characteristic group, the display frames 42 display the moving directions of the moving objects using arrows 43. This allows the user of the external terminal 20 to grasp the traffic volume of moving objects moving straight in each driving lane, as well as the moving directions of the moving objects.

[0095] Although not shown, in the case of the turning system counting result image, similar to the straight-line system counting result image 40, the count values ​​of the moving objects belonging to each turning system similar characteristic group are displayed.

[0096] On the other hand, when the travel route image signal is received by the external terminal 20, a travel route image 50 as shown in Fig. 14 is displayed on the display. The travel route Rm generated as described above is displayed on this travel route image 50. This allows the user of the external terminal 20 to understand the travel route of the moving object when going straight and when turning left.

[0097] As described above, according to the analysis device 10 of the moving body analysis system 1 of this embodiment, the movement characteristic information acquisition unit 12 acquires new movement characteristic information representing the movement characteristics of the moving body based on n moving body images 2x, and the similarity determination unit 13 determines whether the new movement characteristic information has a predetermined movement state similarity with the similar characteristic group information stored in the storage unit 15. Then, in the management unit 14, when the new movement characteristic information has a predetermined movement state similarity with the similar characteristic group information, the moving body corresponding to the new movement characteristic information is added to the similar characteristic group, and the similar characteristic group information is updated by reflecting the new movement characteristic information in the similar characteristic group information.

[0098] On the other hand, when the new movement characteristic information does not have a predetermined movement state similarity with the similar characteristic group information, the moving body is set as a moving body belonging to a new similar characteristic group, and the new movement characteristic information is set as new similar characteristic group information. As described above, every time new movement characteristic information is acquired, the moving body corresponding to the new movement characteristic information is added to the similar characteristic group and the similar characteristic group information is updated without requiring manual operation by the user, thereby making it possible to automatically and appropriately analyze the movement state of the moving body while reducing the burden on the user.

[0099] Furthermore, as described above, the movement state determination unit 12a of the movement characteristic information acquisition unit 12 determines that the movement state of the moving object 8 is in a turning state when the distance Dm between the separation point Pm and the movement line segment Lm is greater than a predetermined value Dref, and determines that the movement state of the moving object 8 is in a straight-line state when the distance Dm is equal to or less than the predetermined value Dref. Since this separation point Pm is the midpoint P that is farthest from the movement line segment Lm, the magnitude of the separation point Pm appropriately indicates whether the movement state of the moving object is in a straight-line state or a turning state. Therefore, according to the above method, it is possible to accurately determine whether the movement state of the moving object is in a straight-line state or a turning state.

[0100] Furthermore, as described above, the final movement characteristic information acquisition unit 12b of the movement characteristic information acquisition unit 12 acquires straight-line movement characteristic information (center of gravity G and first azimuth angle θ1) as the movement characteristic information when the movement state of the moving body is a straight-line state, and acquires turning-line movement characteristic information (start point P1, end point Pn, second azimuth angle θ2, turning angle θt, and separation point Pm) as the movement characteristic information when the movement state of the moving body is a turning state. This makes it possible to create similar characteristic group information and similar characteristic groups for moving bodies while distinguishing between moving bodies that move straight through an intersection and moving bodies that turn right or left.

[0101] Furthermore, when the new movement characteristic information is straight-line movement characteristic information, if all of the above-mentioned conditions (f1) to (f2) are met, it is determined that the new movement characteristic information has a predetermined movement state similarity to the similar characteristic group information, and otherwise it is determined that the new movement characteristic information does not have a predetermined movement state similarity to the similar characteristic group information.

[0102] On the other hand, when the new movement characteristic information is turning-type movement characteristic information, if all of the above-mentioned conditions (f3) to (f6) are met, it is determined that the new movement characteristic information has a predetermined movement state similarity with the similar characteristic group information, and otherwise it is determined that the new movement characteristic information does not have a predetermined movement state similarity with the similar characteristic group information.By using the above method, it is possible to appropriately determine whether the new movement characteristic information has a predetermined movement state similarity with the similar characteristic group information in both cases where the new movement characteristic information is straight-line movement characteristic information and where the new movement characteristic information is turning-type movement characteristic information.

[0103] Furthermore, in the management unit 14, if the new movement characteristic information of the straight-line movement characteristic information has a predetermined similarity in movement state with the straight-line similar characteristic group information, the moving body corresponding to the new movement characteristic information is added to the straight-line similar characteristic group, and as described above, the new movement characteristic information is reflected in the similar characteristic group information, thereby updating the average center of gravity Gave and the first average azimuth angle θ1ave, which are the straight-line similar characteristic group information.

[0104] On the other hand, if the new movement characteristic information of the turning system movement characteristic information has a predetermined movement state similarity with the turning system similar characteristic group information, the moving body corresponding to the new movement characteristic information is added to the turning system similar characteristic group, and as described above, the new movement characteristic information is reflected in the similar characteristic group information, thereby updating the turning system similar characteristic group information, which includes the average start point P1ave, the second average azimuth angle θ2ave, the plane departure point Pmave, and the average end point Pnave.By using the above method, the similar characteristic group information can be calculated as information that reflects the new movement characteristic information in the original similar characteristic group information, and the similar characteristic group information can be appropriately updated.

[0105] Furthermore, the counting unit 16 counts the number of moving objects belonging to the similar characteristic group, the moving route generating unit 17 generates a moving route Rm using the above-mentioned method, and the output unit 18 transmits either a counting result image signal or a moving route image signal to the external terminal 20. When the external terminal 20 receives this counting result image signal, the user U of the external terminal 20 can grasp the traffic volume of moving objects traveling straight in each travel lane and the traffic volume of moving objects turning right or left. On the other hand, when the external terminal 20 receives the moving route image signal, the user U of the external terminal 20 can set security areas along the traveling route of the moving object and determine the direction of entry of the moving object into these security areas. Furthermore, if a crosswalk is present at the intersection, the possibility that the moving object will enter the crosswalk after turning left and the degree of approach of the moving object to the crosswalk can be determined.

[0106] In the embodiment, the midpoint P of the lower side of the bounding box 30 is set as the predetermined point of the moving body image, but instead, the midpoint of any of the right side, left side, or upper side of the bounding box 30, or a vertex of the bounding box, etc. may be set as the predetermined point of the moving body image.

[0107] Furthermore, instance segmentation may be used as the predetermined image recognition process, and in that case, the center of gravity of the moving object may be used as the predetermined point of the moving object image.

[0108] Furthermore, the embodiment is an example in which the movement state of the moving body is determined to be either a straight-line state or a turning state by comparing the distance Dm between the separation point Pm and the movement line segment Lm with a predetermined value Dref. Alternatively, the apex angle of the triangle formed between the separation point Pm and the movement line segment Lm, with the separation point Pm as the vertex, can be compared with a predetermined value, and if this apex angle is smaller than the predetermined value, the movement state of the moving body can be determined to be a turning state, and otherwise the movement state of the moving body can be determined to be a straight-line state.

[0109] Furthermore, although the embodiment is an example in which the mobile object analysis system 1 is applied to a crossroad-shaped intersection 5, the mobile object analysis system of the present invention is not limited to this and can be applied to intersections of various shapes. For example, the mobile object analysis system of the present invention may be applied to a four-way intersection where one of two roads intersects with the other at an angle, a T-shaped or "T"-shaped three-way intersection, or an intersection where five or more roads intersect, such as a five-way intersection.

[0110] On the other hand, although the embodiment is an example in which the intersection is a road intersection, the intersection may instead be an intersection in an indoor passage, in which case the moving object may be a person.

[0111] In addition, the embodiment is an example in which a counting result image signal or a movement route image signal is output from the output unit 18 to the external terminal 20, but instead, the output unit 18 may be configured to output data representing the counting result of the counting unit 16 or the movement route Rm to an output interface such as a printer or display. [Explanation of symbols]

[0112] 1. Mobile object analysis system 2. Traffic environment images 2x moving object images 5 Intersection 8 Mobile 10 Analysis device 11 Moving object image acquisition unit 12. Mobility characteristics information acquisition unit 12a Moving state determination unit 12b Final movement characteristic information acquisition unit 13 Similarity determination section 14 Management Department 15 Storage section 16 Counting section 17 Travel route generation section 17a Driving lane generation unit 17b Intersection generation part 17c 1st point setting section 17d 2nd point setting section 17e Curved path generator 18 Output section G Center of gravity (movement characteristics information, straight-line movement characteristics information) θ1 First azimuth angle (movement characteristic information, straight-line movement characteristic information) P1 Starting point (movement characteristic information, turning movement characteristic information, predetermined point of the oldest moving object image) θ2 Second azimuth angle (movement characteristic information, turning system movement characteristic information) Pn End point (movement characteristic information, turning movement characteristic information, predetermined point of the latest moving object image) Pm Separation point (movement characteristic information, turning system movement characteristic information, first predetermined point) Lb Predetermined reference line Lm moving line segment Gave average center of gravity Ag Predetermined center of gravity range θ1ave 1st average azimuth angle Ld bearing line segment P1ave average starting point θ2ave 2nd average azimuth angle Pmave plane separation point Pnave average end point Ap1 Predetermined starting point range Apm Predetermined distance range Apn Predetermined end range Vp1~Vp4 characteristic vectors L1~L4 driving lanes P13,P24 Approach side intersection P14, P23 Exit side intersection Pin13,Pin24 1st approach point Pout13,Pout24 1st breakaway point Pin14, Pin23 2nd entry point Pout14,Pout23 2nd breakaway point Lc13, Lc24 1st curve Lc14, Lc23 2nd curve

Claims

1. a moving object image acquisition unit that acquires a time series of moving object images, which are images of the moving object, from a traffic environment image that includes a road area including an intersection and a moving object moving in the road area by a predetermined image recognition process; a movement characteristic information acquisition unit that acquires movement characteristic information representing the movement characteristics of the moving object based on the time series of the moving object images; a storage unit that stores similar characteristic groups in which the movement characteristic information is a group of one or more moving objects having a predetermined similarity in movement state, and similar characteristic group information that is data reflecting the movement characteristic information of the one or more moving objects belonging to the similar characteristic group; a similarity determination unit that determines whether or not new movement characteristic information, which is the newly acquired movement characteristic information, has similarity with the similar characteristic group information stored in the storage unit each time the movement characteristic information is newly acquired; a management unit that, when the new movement characteristic information has a similarity to the similar characteristic group information of the predetermined movement state based on the judgment result of the similarity judgment unit, adds the moving body corresponding to the new movement characteristic information to the similar characteristic group, and updates the similar characteristic group information by reflecting the new movement characteristic information in the similar characteristic group information; Equipped with the storage unit stores the similar characteristic group to which the moving entity corresponding to the new movement characteristic information has been added by the management unit and the updated similar characteristic group information as the similar characteristic group and the similar characteristic group information; The mobility characteristic information acquisition unit a moving state determination unit that determines whether the moving state of the moving object is a straight moving state or a turning state based on the time series of moving object images; and a final movement characteristic information acquisition unit that acquires straight-line movement characteristic information as the movement characteristic information when the movement state of the moving body is a straight-line state based on the determination result of the movement state determination unit, and acquires turning-line movement characteristic information as the movement characteristic information when the movement state of the moving body is a turning state. The moving body analysis system is characterized in that, when a line segment connecting a predetermined point on the oldest moving body image in the time series of moving body images and a predetermined point on the latest moving body image in the time series of moving body images is defined as a moving line segment, and a predetermined point on the moving body image between the predetermined point on the oldest moving body image in the time series of moving body images and the predetermined point on the latest moving body image is defined as a first predetermined point, the moving state determination unit determines whether the moving body is in a straight-line state or a turning state based on information on a figure defined by the moving line segment and the first predetermined point.

2. A moving object image acquisition unit that acquires a time series of moving object images, which are images of the moving object, from a traffic environment image that includes a road area including an intersection and a moving object moving in the road area by a predetermined image recognition process; a movement characteristic information acquisition unit that acquires movement characteristic information representing the movement characteristics of the moving object based on the time series of the moving object images; a storage unit that stores similar characteristic groups in which the movement characteristic information is a group of one or more moving objects having a predetermined similarity in movement state, and similar characteristic group information that is data reflecting the movement characteristic information of the one or more moving objects belonging to the similar characteristic group; a similarity determination unit that determines whether or not new movement characteristic information, which is the newly acquired movement characteristic information, has similarity with the similar characteristic group information stored in the storage unit each time the movement characteristic information is newly acquired; a management unit that, when the new movement characteristic information has a similarity to the similar characteristic group information of the predetermined movement state based on the judgment result of the similarity judgment unit, adds the moving body corresponding to the new movement characteristic information to the similar characteristic group, and updates the similar characteristic group information by reflecting the new movement characteristic information in the similar characteristic group information; Equipped with the storage unit stores the similar characteristic group to which the moving entity corresponding to the new movement characteristic information has been added by the management unit and the updated similar characteristic group information as the similar characteristic group and the similar characteristic group information; The mobility characteristic information acquisition unit a moving state determination unit that determines whether the moving state of the moving object is a straight moving state or a turning state based on the time series of moving object images; a final movement characteristic information acquisition unit that acquires straight-line movement characteristic information as the movement characteristic information when the movement state of the moving body is a straight-line state based on the determination result of the movement state determination unit, and acquires turning-line movement characteristic information as the movement characteristic information when the movement state of the moving body is a turning state, the final movement characteristic information acquisition unit acquires, as the straight-line movement characteristic information, a first azimuth angle, which is the angle of a movement line segment, which is a line segment connecting a start point, which is a predetermined point of the oldest moving body image in the time series of the moving body images, and an end point, which is a predetermined point of the latest moving body image in the time series of the moving body images, with respect to a predetermined reference line, and a center of gravity, which is the midpoint between the start point and the end point; the storage unit stores, as the similar characteristic group information, an average center of gravity that is an average position of the center of gravity of the one or more moving bodies and a first average azimuth angle that is an average value of the first azimuth angle; When the new movement characteristic information is the straight-line movement characteristic information, if the center of gravity of the new movement characteristic information is within a predetermined center of gravity range based on the average center of gravity of the similar characteristic group information, and the first azimuth angle of the new movement characteristic information is within a first predetermined angle range based on the first average azimuth angle of the similar characteristic group information, the similarity determination unit determines that the new movement characteristic information has a similarity to the similar characteristic group information in the predetermined movement state, A mobile body analysis system characterized in that, when the new movement characteristic information is the straight-line movement characteristic information and the new movement characteristic information has a similarity to the similar characteristic group information in the specified movement state, the management unit adds the moving body corresponding to the new movement characteristic information to the similar characteristic group, and calculates, as the similar characteristic group information, the average center of gravity and the first average azimuth angle of the moving bodies belonging to the similar characteristic group to which the moving body corresponding to the new movement characteristic information has been added.

3. A moving object image acquisition unit that acquires a time series of moving object images, which are images of the moving object, from a traffic environment image that includes a road area including an intersection and a moving object moving in the road area by a predetermined image recognition process; a movement characteristic information acquisition unit that acquires movement characteristic information representing the movement characteristics of the moving object based on the time series of the moving object images; a storage unit that stores similar characteristic groups in which the movement characteristic information is a group of one or more moving objects having a predetermined similarity in movement state, and similar characteristic group information that is data reflecting the movement characteristic information of the one or more moving objects belonging to the similar characteristic group; a similarity determination unit that determines whether or not new movement characteristic information, which is the newly acquired movement characteristic information, has similarity with the similar characteristic group information stored in the storage unit each time the movement characteristic information is newly acquired; a management unit that, when the new movement characteristic information has a similarity to the similar characteristic group information of the predetermined movement state based on the judgment result of the similarity judgment unit, adds the moving body corresponding to the new movement characteristic information to the similar characteristic group, and updates the similar characteristic group information by reflecting the new movement characteristic information in the similar characteristic group information; Equipped with the storage unit stores the similar characteristic group to which the moving entity corresponding to the new movement characteristic information has been added by the management unit and the updated similar characteristic group information as the similar characteristic group and the similar characteristic group information; The mobility characteristic information acquisition unit a moving state determination unit that determines whether the moving state of the moving object is a straight moving state or a turning state based on the time series of moving object images; a final movement characteristic information acquisition unit that acquires straight-line movement characteristic information as the movement characteristic information when the movement state of the moving body is a straight-line state based on the determination result of the movement state determination unit, and acquires turning-line movement characteristic information as the movement characteristic information when the movement state of the moving body is a turning state, The final movement characteristic information acquisition unit acquires, as the turning system movement characteristic information, a start point, which is a predetermined point of the oldest moving body image in the time series of the moving body images, a second azimuth angle, which is the angle with respect to a predetermined reference line of an azimuth line, which is a line segment connecting the start point and a predetermined point of the moving body image next to the oldest moving body image, an end point, which is a predetermined point of the newest moving body image in the time series of the moving body images, and a separation point, which is a point among the predetermined points in the time series of the moving body images that is the longest distance from the movement line segment, which is a line segment connecting the start point and the end point, the storage unit stores, as the similar characteristic group information, an average start point which is an average position of the start points of the one or more moving bodies, a second average azimuth angle which is an average value of the second azimuth angles, an average departure point which is an average position of the departure points, and an average end point which is an average position of the end points; When the new movement characteristic information is the turning movement characteristic information, if the following are all true: the starting point in the new movement characteristic information is within a predetermined starting point range based on the average starting point of the similar characteristic group information; the second azimuth angle of the new movement characteristic information is within a second predetermined angle range based on the second average azimuth angle of the similar characteristic group information; the departure point of the new movement characteristic information is within a predetermined departure point range based on the average departure point of the similar characteristic group information; and the end point of the new movement characteristic information is within a predetermined end point range based on the average end point of the similar characteristic group information, the similarity determination unit determines that the new movement characteristic information has similarity to the predetermined movement state of the similar characteristic group information, The management unit, when the new movement characteristic information is the turning-type movement characteristic information and the new movement characteristic information has a similarity to the similar characteristic group information in the specified movement state, adds the moving body corresponding to the new movement characteristic information to the similar characteristic group, and calculates, as the similar characteristic group information, the average starting point, the second average azimuth angle, the average separation point, and the average ending point of the moving bodies belonging to the similar characteristic group to which the moving body corresponding to the new movement characteristic information has been added.

4. 3. The mobile object analysis system according to claim 1, When the new movement characteristic information does not have similarity with the similar characteristic group information, the management unit sets the moving body corresponding to the new movement characteristic information as a moving body belonging to a new similar characteristic group, and sets the new movement characteristic information as new similar characteristic group information; the storage unit further stores the new similar characteristic group and the new similar characteristic group information in addition to the similar characteristic group and the similar characteristic group information; A mobile object analysis system characterized in that the similarity determination unit determines, each time new movement characteristic information is acquired, whether the new movement characteristic information has similarity in the specified movement state with the similar characteristic group information stored in the memory unit and the new similar characteristic group information.

5. The moving body analysis system according to any one of claims 1 to 3, a counting unit that counts the number of the one or more moving objects that belong to the similar characteristic group; A mobile object analysis system further comprising: an output unit that outputs count information representing the number of the one or more mobile objects counted by the counting unit.

6. 3. The mobile object analysis system according to claim 2, the storage unit further stores, as the similar characteristic group information, a characteristic vector extending from the mean centroid along the first mean azimuth angle, a driving lane generation unit that generates, when a plurality of pieces of similar characteristic group information are stored in the storage unit, a straight line that passes through the average center of gravity in each of the plurality of pieces of similar characteristic group information and extends at the first average azimuth angle as a driving lane; an intersection point generating unit that generates an intersection point between two of the plurality of driving lanes when the intersection angle between the two driving lanes is within a range of 90°±γ (γ is a positive value); a first point setting unit that, when the two intersections are generated on one of the driving lanes, sets one of the two intersections as an entrance-side intersection on the side where the moving body enters and sets the other of the two intersections as a departure-side intersection on the side where the moving body leaves, based on the similar characteristic group information corresponding to the one of the driving lanes; a second point setting unit that sets a point on the one driving lane that is spaced a predetermined value from the entry-side intersection on the opposite side of the exit-side intersection as a first entry point, sets a point on a driving lane that intersects with the one driving lane at the entry-side intersection in the direction of the characteristic vector by the predetermined value, sets a point on the driving lane that intersects with the one driving lane at the exit-side intersection on the opposite side of the characteristic vector by the predetermined value as a second entry point, and sets a point on the one driving lane that is spaced a predetermined value from the exit-side intersection on the opposite side of the entry-side intersection as a second departure point; a curved path generating unit that generates a first curved path connecting the first entry point and the first departure point as a parametric curved path having a maximum curvature at a portion closest to the entry-side intersection, and that generates a second curved path connecting the second entry point and the second departure point as a parametric curved path having a maximum curvature at a portion closest to the departure-side intersection; A moving object analysis system further comprising:

Images