JP7256032B2 - Person counting device, person counting method and person counting program - Google Patents

Description

The present invention relates to a person counting device, a person counting method, and a person counting program. More specifically, the present invention relates to a person counting device, a person counting method, and a person counting program suitable for counting the number of persons who have passed through a counting area based on an image of a predetermined area.

2. Description of the Related Art Conventionally, there has been proposed a method of analyzing a video of a predetermined indoor/outdoor area captured by a surveillance camera and counting the number of people who have passed through the predetermined area.

For example, Patent Literature 1 discloses a method of setting a line segment in an image as a reference for the entry and exit of a person, and counting the person when the movement vector of each pedestrian crosses the line segment. there is

In addition, in Patent Document 2, in order to analyze the movement of customers in the store, the movement pattern of the person is specified based on the intersection of each side of the measurement area and the flow line of the person, A method for counting the number of people for each movement pattern is disclosed.

JP 2016-71801 A Patent No. 5834249

However, in the method described in Patent Literature 1, when a plurality of people overlap during congestion, there is a possibility that the motion vector is interrupted before and after the line segment (counting line), resulting in an omission in counting.

In the method described in Patent Document 2, the object to be measured is an area (plane), but it is necessary to detect the intersection of at least one side of the measurement area and the person's flow line. must cross at least one side of Therefore, even with this method, when a plurality of people overlap during congestion, there is a possibility that the line of flow will be interrupted before and after the side of the measurement area, and that person will be omitted from the count.

In view of the above problems, the present invention provides a people counting device, a people counting method, and a people counting program capable of reducing omissions in counting the number of people even during congestion and improving the accuracy of counting people. With the goal.

In order to solve the above-described problems and achieve the object, the present invention provides a person counting device that counts the number of persons who have passed through a counting area within a predetermined area based on a video image of the predetermined area. a trajectory detection unit for detecting a movement trajectory for each person from the video; and a person counting unit that counts the number of persons who have passed through the counting area based on the number of persons.

Further, according to the present invention, in the above invention, the movement trajectory having a length of a predetermined length or more in the counting area is present in the counting area for a predetermined time or longer. .

Further, in the above invention, the moving trajectory whose length in the counting area is equal to or greater than a predetermined length has a number of frames existing in the counting area equal to or greater than a predetermined number of frames. and

Further, according to the present invention, in the above invention, the moving trajectory having a length in the counting area of a predetermined length or more has a spatial length in the counting area of a predetermined spatial length or more. characterized by being

Further, according to the present invention, in the above invention, the person counting unit has a moving trajectory detected by the trajectory detecting unit whose length in the counting area is equal to or greater than a predetermined length, and whose moving direction is a predetermined length. The number of persons who have passed through the counting area is counted based on the number of movement trajectories in the direction of .

Further, according to the present invention, in the above invention, the trajectory detection section detects a movement trajectory of each person from the video within the counting area.

Further, according to the present invention, in the above invention, the trajectory detection section detects a movement trajectory of each person from the image in the entire predetermined area.

Further, in the present invention, in the above invention, the person counting unit calculates the number of movement trajectories detected by the trajectory detection unit, excluding a movement trajectory having an average movement amount smaller than a predetermined value in the counting area. The number of persons who have passed through the counting area is counted based on the number of lines.

Further, according to the present invention, in the above-described invention, the counting area is a polygon, and the person counting section moves from one side of the counting area to the same side of the moving trajectory detected by the trajectory detecting section. The number of persons who have passed through the counting area is counted based on the number of movement trajectories excluding the movement trajectories returning to .

Further, in the present invention, in the above-described invention, the person counting unit includes a movement trajectory detected by the trajectory detection unit excluding a movement trajectory having an end point portion in a moving direction opposite to a starting point portion moving direction. The number of persons passing through the counting area is counted based on the number of tracks.

Further, according to the present invention, in the above invention, the trajectory detection unit detects at least two of a part on the central axis of the person, a part off to the left from the central axis, and a part off to the right from the central axis. It is characterized by detecting the movement trajectory of each person by tracking each of the six parts.

Further, the present invention is a person counting method for counting the number of persons who have passed through a counting area within a predetermined area based on an image of a predetermined area, wherein the movement trajectory of each person is detected from the image. a trajectory detection step; and the number of people who have passed through the counting area based on the number of trajectories detected by the trajectory detection step that have a length equal to or greater than a predetermined length in the counting area. and a person counting step of counting

Further, the present invention is a person counting program for counting the number of persons who have passed through a counting area within a predetermined area based on an image of a predetermined area, wherein the moving trajectory of each person is calculated from the image. A person who has passed through the counting area based on the trajectory detection process to be detected and the number of trajectories among the trajectories detected by the trajectory detection process whose length in the counting area is equal to or greater than a predetermined length. and a person counting process for counting the number of people.

According to the people counting device, the people counting method, and the people counting program of the present invention, it is possible to reduce omissions in counting the number of people even during congestion, and improve the accuracy of counting people.

FIG. 10 is a diagram for explaining how a person's movement trajectory is detected in a counting area set in an image; FIG. 4 is a diagram for explaining the detected movement trajectories of each person and the number of frames in which they are detected within the counting area; 1 is a block diagram illustrating the overall configuration of a system including a people counting device according to Embodiment 1; FIG. 1 is a block diagram illustrating the configuration of a person counting device according to Embodiment 1; FIG. FIG. 4 is a diagram for explaining body parts tracked by a trajectory detection unit; FIG. 10 is a diagram for explaining a method of determining from which side of a counting area a moving trajectory starts and reaches which side of a counting area; 6 is a flow chart showing an example of a procedure of people counting processing performed by the people counting device according to the first embodiment; It is a schematic diagram which shows an example of the counting result display screen displayed on a monitor. FIG. 11 is a schematic diagram showing another example of the counting result display screen displayed on the monitor; FIG. 10 is a schematic diagram showing still another example of the counting result display screen displayed on the monitor; FIG. 4 is a diagram for explaining the movement trajectory of each detected person and their spatial length within the counting area;

In this specification, the length in the description that "the length in the counting area is a predetermined length or more" refers to the temporal length (time) and the spatial length (trajectory length) shall include both Therefore, "the length in the counting area is a predetermined length or longer" means (1) the time in the counting area, that is, the time for which the trajectory exists in the counting area is a predetermined time or longer; (2) It means that at least one of the following conditions is satisfied: that the spatial length (length of the trajectory) in the counting area is equal to or greater than a predetermined spatial length.

In addition, the time in the case of (1) above does not only directly represent the length of time such as seconds, but also indirectly represents the length of time such as the number of frames of video (video). shall also be included.

BEST MODE FOR CARRYING OUT THE INVENTION Preferred embodiments of a person counting device, a person counting method, and a person counting program according to the present invention will be described below with reference to the drawings.

<Overview of this embodiment>
First, an overview of the present embodiment will be described with reference to FIGS. 1 and 2. FIG. In this embodiment, first, as shown in FIG. 1, a polygonal (for example, square) counting area having a certain area is set in an image of a predetermined area (for example, an outdoor passage). Next, each person passing through the counting area is tracked to detect the movement trajectory (hereinafter simply trajectory) of each person. Then, the time in the counting area (the time in the counting area), here, the number of trajectories where the number of frames f existing in the counting area is equal to or greater than a predetermined number of frames (hereinafter referred to as the reference number of frames) is counted. As a result, even if the trajectory of a person is interrupted during congestion, the person can be counted. That is, as shown in FIG. 2, unlike the methods described in Patent Documents 1 and 2, in the present embodiment, even if the trajectory A does not intersect the counting area, the number of frames f existing within the counting area is used as the reference. If the number of frames is greater than or equal to the number of frames, the person can be counted. Therefore, it is possible to reduce the occurrence of counting omissions due to discontinuities in the trajectory.

Of course, even if the trajectory B or the trajectory C crosses the counting area, the person can be counted if the number of frames f present in the counting area is equal to or greater than the reference number of frames.

On the other hand, for the trajectory D whose frame number f is less than the reference frame number, the person is excluded from the counting target. This is to exclude trajectories that are erroneously detected for a short period of time, such as erroneously detecting a person other than a person, such as a bicycle, from the objects to be counted. Such erroneous detection is effectively eliminated by setting the number of reference frames to N (N is an integer equal to or greater than 2) and excluding trajectories for which only (N-1) frames have been detected. be able to.

As described above, in this embodiment, it is possible to reduce omissions in counting the number of people even during congestion, and to improve the accuracy of counting the number of people.

Further, in this embodiment, as shown in FIG. 2, when there are trajectories moving in various directions, the trajectories are moved in predetermined directions (for example, upward and downward directions as shown in FIG. 2). (2 directions) and counting, and the number of persons can be counted for each predetermined direction.

In addition, in the present embodiment, by excluding the following cases from the counting target, the counting accuracy of persons passing through the counting area can be increased.

1) When the average amount of movement in the trajectory counting area is smaller than a predetermined value.
This makes it possible to exclude persons staying in the counting area for a long period of time from being counted. If a person staying for a long period of time is not excluded from counting, tracking will be interrupted and multiple trajectories will be detected when another person passes in front of the person. You may count the number of times.

2) When the trajectory makes a U-turn.
As a result, a person who has made a U-turn within the counting area can be excluded from being counted. Therefore, it is possible to improve the counting accuracy of persons passing through the counting area.

The present embodiment will be described in more detail below. A case will be described below in which tracking of a person is not performed outside the counting area, but only within the counting area.

<Overall Configuration of System Including People Counting Device>
First, the overall configuration of the system including the people counting device according to the first embodiment will be described with reference to FIG. As shown in FIG. 3, this system is constructed for a predetermined indoor and outdoor area (for example, an outdoor passage), etc., and includes a camera 1, a recorder (image recording device) 2, and a person counting device. 3 and .

The camera 1 is installed at a desired location indoors or outdoors. A predetermined indoor and outdoor area is photographed by the camera 1. - 特許庁The captured video (each frame) is accumulated in the recorder 2 .

The people counting device 3 is connected to a monitor (display device) 4 for displaying captured images, etc., and an input device 5 (for example, a mouse) for an operator to perform various input operations. Note that the monitor 4 and the input device 5 may be composed of a display device with an input function such as a touch panel display.

The people counting device 3 is configured so that the operator can view the video captured by the camera 1 in real time on the monitor 4 and also can view the past video recorded on the recorder 2. there is

<Configuration of person counting device>
Next, the configuration of the person counting device 3 will be described with reference to FIG. The people counting device 3 is an information processing device having functions equivalent to those of a general personal computer, and includes a control unit 10 and a storage unit 20 as shown in FIG.

The control unit 10 has functions of an image input unit 11 , a trajectory detection unit 12 , a counting area setting unit 13 , a person counting unit 14 and a display information generation unit 15 . The control unit 10 includes, for example, software programs for realizing various types of processing, a CPU (Central Processing Unit) that executes the software programs, various types of hardware controlled by the CPU, and the like. Software programs and data necessary for the operation of the control unit 10 are stored in the storage unit 20 .

4 of the control unit 10 are implemented by causing the CPU of the control unit 10 to execute a people counting program. The people counting program may be installed in the people counting device 3 in advance, or may be recorded in a computer-readable recording medium as an application program operable on a general-purpose OS, or sent to the operator via a network. may be provided.

The storage unit 20 is configured by a storage device such as a hard disk device or a nonvolatile memory.

The image input unit 11 performs processing for acquiring images from the camera 1 or the recorder 2 . When the people are counted in real time, the images are obtained from the camera 1, and when the people are counted from past images, the images are obtained from the recorder 2. - 特許庁

The trajectory detection unit 12 detects the trajectory of each person within the counting area from the image captured by the camera 1 . This trajectory detection process can use known image recognition technology (person detection technology, person tracking technology, etc.).

Specifically, for example, the trajectory detection unit 12 first uses an existing person detection technology to detect the head, shoulders, elbows, hands, waist, knees, legs, etc. of each person in the entire predetermined area photographed. The body part is detected to estimate the posture, and the coordinate information of each person is generated for each frame. The coordinate information includes position information of each detected part and information about the time when the detected person was photographed. Note that the photographing time of the detected person can be obtained from the photographing time of the frame.

Subsequently, the generated coordinate information of the person is collated between consecutive frames, and the person having the coordinate information that satisfies the criteria is associated with each other, thereby tracking each person within the counting area. The association is made, for example, by assigning a common ID number to the person. At this time, the trajectory detection unit 12 keeps tracking the person for a predetermined number of frames (for example, M frames, where M is an integer equal to or greater than 2) even if tracking of the person is interrupted because the person is hidden by another person. continue. For example, while holding the coordinate information before the tracking is interrupted, the position of the person after that is estimated for each frame. is the same person as the person before the tracking was interrupted. In addition, in this embodiment, when tracking of a person is interrupted, the number f of frames in which the trajectory exists within the counting area is counted by adding the number of frames while tracking is interrupted. It should be noted that the above M frame can be changed as appropriate according to the settings of the camera 1 .

When tracking a person, it is preferable to track at least two parts of the person on the central axis, the part off the central axis to the left, and the part off the central axis to the right. As a result, it is possible to suppress the occurrence of an error in association when two persons pass each other. This is because it is possible to determine that the two persons are not the same person because the left and right parts of the two persons passing each other are positioned opposite to each other with respect to the central axis. For example, as shown in FIG. 5, the center of the left and right shoulders on the central axis of the person, the left shoulder and left hip off the central axis to the left, and the right shoulder and right hip off the central axis to the right, respectively. Chase. Then, the movement distance of each part between frames is calculated, the average of the calculated movement distances of the five parts is taken, and the person is associated based on the average distance. Also, for example, the part on the central axis of the person and the right shoulder may be tracked, or the part on the central axis of the person and the left shoulder may be tracked. It should be noted that, even with the existing person detection technology, it is possible to distinguish between the left and right parts of a person and detect them, so that the left and right parts can be distinguished and easily tracked as described above.

Then, the trajectory detection unit 12 performs processing for generating the trajectory of the person based on the associated coordinate information of the person. A trajectory is generated by connecting specific points that detect specific parts (eg, the head).

Trajectory information 21 regarding the trajectory of each person detected by the trajectory detection unit 12 is stored in the storage unit 20 by the trajectory detection unit 12 .

The counting area setting unit 13 performs a process of setting a polygonal (for example, quadrangular) counting area having a certain area in accordance with an operator's input operation. The operator uses the input device 5 such as a mouse to input the position of the counting area on the screen of the monitor 4 on which the image obtained from the camera 1 or the recorder 2 is displayed.

The person counting unit 14 counts the number of persons who have passed through the counting area based on the number of trajectories for each person included in the trajectory information 21 acquired by the trajectory detection unit 12 . In this embodiment, when the number of frames in which the trajectory is detected in the counting area (=frame number f) is greater than or equal to the reference number of frames, it is determined that the person has passed through the counting area, and the person is counted. .

In addition, as shown in FIG. 2, when there are a plurality of trajectories moving in various directions, the person counting unit 14 detects the trajectories in each predetermined direction (for example, in two directions, upward and downward). ) and counted. Thereby, the number of persons can be counted for each predetermined direction.

It should be noted that the direction in which the trajectory is to be classified may be determined by comparing the coordinates of the starting point (starting point) and the end point of the trajectory. For example, in the case shown in FIG. 6, if the Y coordinate of the start point of the locus is smaller than the Y coordinate of the end point of the locus, it is determined that the direction is upward. If it is not smaller, it may be determined to be downward. Also, if there is a difference in the horizontal coordinates of the start point and the end point of the trajectory, the left and right directions may be counted at the same time. That is, if the X coordinate of the start point of the locus is smaller than the X coordinate of the end point of the locus, it is determined to be in the rightward direction. can be determined.

In addition, when the trajectory remains within the counting area, specifically, when the average amount of movement of the trajectory within the counting area is smaller than a predetermined value, the person counting unit 14 excludes the trajectory from being counted. . This makes it possible to exclude persons staying in the counting area for a long period of time from being counted. The average amount of movement may be, for example, the average amount of movement per predetermined number of frames (for example, one frame), or the average amount of movement per predetermined time period (for example, one second).

Furthermore, when the trajectory makes a U-turn within the counting area, the person counting unit 14 also excludes that trajectory from being counted. As a result, a person who has made a U-turn within the counting area can be excluded from being counted.

More specifically, the person counting unit 14 excludes the trajectory from counting when the trajectory progresses from a side of the counting area into the counting area and then returns to the same side.

As shown in FIG. 6, from which side of the counting area the trajectory advances into the counting area is determined by the first predetermined number of frames (for example, P frames, where P is an integer of 2 or more) where the trajectory is drawn. is used to calculate the average direction of travel, and is determined by finding the edges that intersect by extending the trajectory from the start of the trajectory in the opposite direction to the direction of travel. Similarly, to which side of the counting area the trajectory returns (progresses) is determined by using the last predetermined number of frames (for example, Q frames, where Q is the same number as P) to draw the trajectory, and the average moving direction , and extending the trajectory from the endpoint of the trajectory in the direction of its movement and detecting the intersecting edges.

The person counting unit 14 may exclude the trajectory from being counted when the direction of movement of the end point of the trajectory is opposite to the direction of movement of the starting point of the trajectory. As a result, even if the counting area is not a polygon such as a circle or an ellipse, it is possible to effectively exclude a person who makes a U-turn within the counting area from being counted. It should be noted that whether or not the movement directions of the end point portion and the start point portion of the trajectory are opposite to each other can be determined by calculating the average movement direction of the end point portion and the start point portion as shown in FIG. Compared with an angle range (for example, a range of 180°±20°), if it is within the predetermined angle range, it is determined to be opposite, and if it is not within the predetermined angle range, it is determined to be non-opposite.

The display information generation unit 15 performs processing for generating display information representing the number of persons who have passed through the counting area based on the counting result of the person counting unit 14 . The display information generation unit 15 also performs processing for generating display information representing the trajectory of a person passing through the counting area based on the trajectory information 21 regarding the trajectory of each person detected by the trajectory detection unit 12 . The generated display information is output to the monitor 4, and the number of persons who have passed through the counting area is displayed in each predetermined direction on the monitor 4, and the trajectory of each person is displayed.

<Procedure for counting people>
Next, the procedure of the people counting process performed by the people counting device 3 will be described with reference to FIG.

As shown in FIG. 7, first, a video (frame) is input to the video input unit 11 from the camera 1 or the recorder 2 (S11).

Next, the trajectory detection unit 12 detects the trajectory of the person from the input video (S12). Note that when a plurality of persons are detected from the video, the trajectory detection processing by the trajectory detection unit 12 and the subsequent processing are performed for each person.

Next, the person counting unit 14 performs the following processing. First, it is determined whether or not the number f of frames in which the detected trajectory exists within the counting area is greater than or equal to the reference number of frames (S13). More specifically, whether or not the number of frames in which the trajectory is tracked within the counting area (including the number of frames during which tracking is temporarily interrupted) is N frames (N is an integer of 2 or more) or more. If the number of frames tracked in the counting area is less than N frames (No in S13), the people counting process is terminated without counting the trajectory (S14). If the number of frames tracked within the counting area is N frames or more (Yes in S13), it is determined whether or not the trajectory remains within the counting area (S15). If the trajectory stays within the counting area (Yes in S15), the people counting process ends without counting the trajectory (S14). If the trajectory does not stay within the counting area (No in S15), it is determined whether or not the trajectory makes a U-turn within the counting area (S16). If the trajectory makes a U-turn within the counting area (Yes in S16), the people counting process is terminated without counting the trajectory (S14). If the trajectory does not make a U-turn within the counting area (No in S16), the Y coordinate of the starting point of the trajectory (the Y coordinate in the first frame in which the trajectory is detected) is the Y coordinate of the ending point of the trajectory (the trajectory is It is determined whether or not it is smaller than the detected Y coordinate in the final frame (S17). If the Y-coordinate locus of the starting point of the locus is not smaller than the Y-coordinate of the ending point of the locus (No in S17), processing is performed to increment the count value in the upward direction by one (S18). If the Y-coordinate locus of the starting point of the locus is smaller than the Y-coordinate of the ending point of the locus (Yes in S17), the downward count value is incremented by one (S19).

In the example shown in FIG. 7, the number of trajectories is counted for each predetermined direction, but this processing may be performed regardless of the direction of the trajectory. That is, if the trajectory does not make a U-turn within the counting area (No in S16), the count value may be immediately incremented by one.

<Browsing screen for people counting results>
Next, the people counting screen for the operator to view the people counting results will be described with reference to FIGS.

In this embodiment, as shown in FIG. 8, the display information generating unit 15 generates display information representing the number of persons who have passed through the counting area based on the counting result of the person counting unit 14, and displays the information on the monitor 4. A count result display screen is displayed on the monitor 4 for the operator to view the count result. A video display section 41 and a counting result display section 42 are provided on the counting result display screen.

The video display unit 41 overlays the counting area and the trajectory of each person with the video from the camera 1 . The number of people passing through the counting area, which is the result of counting the number of people, is overlaid on the counting result display section 42 for each predetermined direction (for example, upward and downward directions).

FIG. 9 is an example of a counting result display screen showing a case where the counting area is a polygon other than a quadrangle, specifically a pentagon.

FIG. 10 is an example of a count result display screen showing the result of counting the number of people entering and leaving an indoor room. Is displayed.

In this example, a hexagonal counting area with six vertices is set. Each vertex can be freely moved by operating (clicking and dragging) the mouse, which is the input device 5, and the operator can easily set the counting area at a desired position.

Also, in the center of the counting area, a double-headed arrow indicating the direction of entry (IN) and the direction of exit (OUT) is displayed. This double-headed arrow can be rotated by operating the mouse. For example, clicking the center and dragging it to the right rotates it to the right, and dragging it to the left rotates it to the left. Thereby, the operator can easily set the direction of entering (IN) and the direction of leaving (OUT).

As described above, in the present embodiment, the time in the counting area, more specifically, the number of trajectories for which the number of frames f existing in the counting area is equal to or greater than a predetermined number of frames (reference number of frames) is counted. Since the number of people who have passed through the area is counted, even if the trajectory of a person is interrupted during congestion, the person can be counted. Therefore, it is possible to reduce omissions in counting the number of people even during congestion, and improve the accuracy of counting the number of people.

In the above-described embodiment, the number of people who have passed through the counting area is counted based on the number of trajectories whose temporal length (time) is equal to or greater than a predetermined time in the counting area. As shown in FIG. 11, based on the number of trajectories whose spatial length (trajectory length) L in the counting area is equal to or greater than a predetermined spatial length (hereinafter referred to as reference length), counting The number of persons passing through the area may be counted. In this case also, even if the trajectory of a person is interrupted during congestion, the person can be counted. That is, as shown in FIG. 11, even if the trajectory a does not intersect the counting area, if the length L within the counting area is equal to or greater than the reference length, the person can be counted. Therefore, it is possible to reduce the occurrence of counting omissions due to discontinuities in the trajectory.

Of course, even if the trajectory b or trajectory c intersects the counting area, if the length L in the counting area is equal to or greater than the reference length, the person can be counted. On the other hand, from the viewpoint of excluding trajectories that are erroneously detected from being counted, for trajectories d whose length L is less than the reference length, the person is excluded from being counted.

Also, the number of people who have passed through the counting area is counted based on the number of trajectories that exist in the counting area for a predetermined time or longer and whose length L in the counting area is a reference length or longer. You may

The spatial length (trajectory length) in the counting area may be calculated from the coordinates (pixels) of all specific points in the counting area, or may be first detected in the counting area. It may be calculated from the coordinates (pixels) of the detected specific point, that is, the start point, and the coordinates (pixels) of the last detected specific point, that is, the end point within the counting area.

Further, in the above-described embodiment, the case where the tracking of a person is performed only within the counting area without being performed outside the counting area has been mainly described. Alternatively, the trajectory of each person may be detected in the entire predetermined area.

Although the embodiments of the present invention have been described above with reference to the drawings, the present invention is not limited to the above embodiments. Moreover, the configuration of each embodiment may be appropriately combined or changed without departing from the gist of the present invention.

INDUSTRIAL APPLICABILITY As described above, the present invention is a technique useful for counting the number of persons who have passed through a counting area within a predetermined area based on a video image of the predetermined area.

1: Camera 2: Recorder 3: People counting device 4: Monitor 5: Input device 10: Control unit 11: Video input unit 12: Trajectory detection unit 13: Counting area setting unit 14: People counting unit 15: Display information generation unit 20 : Storage unit 21: Trajectory information 41: Image display unit 42: Count result display unit

Claims (12)

A people counting device that counts the number of people who have passed through a counting area within a predetermined area based on an image of the predetermined area,
a trajectory detection unit that detects a movement trajectory including coordinate information for each person from the video by connecting specific points that detect specific parts of the person ;
calculating the spatial length of the movement trajectory within the counting area from the coordinates of the specific point within the counting area ; a person counting unit that counts the number of movement trajectories whose spatial length is equal to or greater than a predetermined spatial length, and counts the number of persons who have passed through the counting area based on the counted number of the movement trajectories; ,
A person counting device comprising: 2. The movement trajectory whose spatial length in said counting area is equal to or longer than a predetermined spatial length is present in said counting area for a time equal to or longer than a predetermined time. A person counting device as described. The moving trajectory having the spatial length within the counting area equal to or greater than a predetermined spatial length has a number of frames existing within the counting area equal to or greater than a predetermined number of frames. Item 3. The person counting device according to item 1 or 2. The person counting unit determines that the spatial length in the counting area of the moving trajectory detected by the trajectory detecting unit is equal to or greater than a predetermined spatial length and the moving direction is a predetermined direction. 4. The person counting device according to any one of claims 1 to 3 , wherein the number of persons passing through said counting area is counted based on the number of movement trajectories. 5. The person counting device according to any one of claims 1 to 4 , wherein the trajectory detection unit detects a movement trajectory of each person from the video within the counting area. 5. The person counting device according to any one of claims 1 to 4 , wherein the trajectory detection unit detects a movement trajectory of each person from the image in the entire predetermined area. The person counting unit passes through the counting area based on the number of trajectories detected by the trajectory detecting unit, excluding trajectories whose average amount of movement in the counting area is smaller than a predetermined value. 7. The person counting device according to any one of claims 1 to 6 , which counts the number of persons who have done so. the counting area is a polygon,
The person counting unit passes through the counting area based on the number of moving trajectories detected by the trajectory detecting unit, excluding moving trajectories that proceed from one side of the counting area and return to the same side. 8. The person counting device according to any one of claims 1 to 7 , which counts the number of persons who have done so. The person counting section calculates the counting area based on the number of movement trajectories detected by the trajectory detection section, excluding a movement trajectory having an end portion whose movement direction is opposite to a movement direction of a start point portion. 9. The person counting device according to any one of claims 1 to 8 , which counts the number of persons who have passed through. The trajectory detection unit tracks at least two parts of a part on the central axis of the person, a part off to the left from the central axis, and a part off to the right from the central axis. Calculate the movement distance between frames, take the average of the calculated movement distances of a plurality of parts, associate the person based on the average distance, and based on the coordinate information of the associated person, 10. The person counting device according to any one of claims 1 to 9, wherein the movement trajectory of is generated by connecting specific points from which specific parts are detected . A person counting method for counting the number of persons who have passed through a counting area within a predetermined area based on a video image of the predetermined area,
a trajectory detection step of detecting a movement trajectory including coordinate information for each person from the video by connecting specific points from which specific parts of the person are detected ;
calculating the spatial length of the movement trajectory within the counting area from the coordinates of the specific point within the counting area ; a person counting step of counting the number of movement trajectories having a spatial length equal to or greater than a predetermined spatial length, and counting the number of persons who have passed through the counting area based on the counted number of the movement trajectories; ,
A person counting method, comprising: A person counting program for counting the number of persons who have passed through a counting area within a predetermined area based on a video image of the predetermined area,
A trajectory detection process for detecting a movement trajectory containing coordinate information for each person from the video by connecting specific points from which specific parts of the person are detected ;
calculating the spatial length of the movement trajectory within the counting area from the coordinates of the specific point within the counting area ; a person counting process for counting the number of movement trajectories having a spatial length equal to or greater than a predetermined spatial length, and counting the number of persons who have passed through the counting area based on the counted number of the movement trajectories; ,
A person counting program characterized by causing a computer to execute

Images