JP2024040800A - Passing person measuring device - Google Patents

Abstract

[Problem] To provide a passing people counting device that is less likely to cause erroneous measurement than conventional devices. [Solution] A passing people counting device 10A includes a tracking target recognition unit 22 that recognizes a tracking target based on an area image generated every time Δt passes, an occurrence position memory unit 24 that stores to which of a plurality of divided areas that constitute the area image the occurrence position of the recognized tracking target belongs, a position prediction unit 25 that predicts the position at time t+Δt of a tracking target included in a current area image that is an area image generated at time t, and a passing processing unit 26 that, when a tracking target whose predicted position at time t+Δt belongs to an outer edge area of a divided area different from the divided area to which the occurrence position belongs is included in the current area image, counts up the number of passing people by the number of tracking targets and further sets a prohibited area around the position of the tracking target at time t+Δt. [Selected Figure] Figure 1

Description

The present invention relates to a passing person measuring device that measures the number of people passing through a detection area provided near the entrances and exits of various facilities and various vehicles.

BACKGROUND ART Conventionally, a room entry/exit monitoring device described in Patent Document 1 is known. This monitoring device consists of a pair of heat sensing sensors (an outdoor thermopile sensor and an indoor thermopile sensor) installed above the entrance, an amplifier that amplifies the difference between the output signals of each sensor, and a It also includes a CPU processing unit that determines human movements. This monitoring device can detect not only when a person enters or leaves a room, but also when a person who attempts to enter the room makes a U-turn without entering the room, or when a person who attempts to leave the room makes a U-turn without leaving the room. I can do it. However, this monitoring device cannot accurately identify the number of people entering or leaving the room at the same time. The same applies when entering and exiting the room occur at the same time.

To solve this problem, we use a thermal image sensor that generates thermal images of the detection area near entrances and exits at regular intervals, and identify human movements in the detection area by analyzing the generated thermal images. A type of monitoring device equipped with an image analysis section has been proposed. According to this monitoring device, it is possible to specify the passing direction and number of people passing through the detection area. For example, with this monitoring device, it is possible to identify that two people have entered through the doorway and that three people have left the doorway.

By the way, in a type of monitoring device that uses thermal images, the detection area is detected by tracking clusters of pixels that indicate thermal changes in thermal images generated at predetermined intervals. It is common to identify people's movements. However, with this method, it is difficult to identify the movement of a person approaching the outer edge of the detection area, and as a result of failing to identify the movement of a person, the number of people passing by may be incorrectly measured.

Japanese Patent Application Publication No. 9-113355

The present invention has been made in view of the above-mentioned circumstances, and an object of the present invention is to provide a passing person measuring device that is less likely to cause erroneous measurements than conventional ones.

In order to solve the above problems, a passing person measuring device according to the present invention is a device that measures the number of people passing through a predetermined detection area, and includes an area related to a detection area that is generated every time Δt elapses. A tracking target recognition unit that recognizes a tracking target corresponding to a person based on the image, an identification number assigning unit that assigns an identification number to the recognized tracking target, and the location of the recognized tracking target constitute an area image. an occurrence position storage unit that stores which of a plurality of divided areas the tracking target belongs to in association with an identification number assigned to the tracking target; a position prediction unit that predicts the position of a tracking target at time t+Δt, and the current area image includes a tracking target whose position at the predicted time t+Δt belongs to an outer edge area of a divided area that is different from the divided area to which the occurrence position belongs. In this case, the identification number is deleted from the tracking target, the number of people is counted up by the number of tracking targets, and a prohibited area is set around the position of the tracking target at time t or time t + Δt. a processing unit, and the identification number assigning unit determines that a tracking target corresponding to the tracking target included in the current area image is included in a previous area image that is an area image generated at time t−Δt. In this case, the tracking target included in the current area image is given the same identification number as the tracking target included in the previous area image, while the tracking target included in the current area image is assigned the same identification number as the tracking target included in the current area image. If it is determined that the tracking target is not included in the previous area image, a new identification number is assigned to the tracking target included in the current area image, and the identification number assigning unit assigns a new identification number to the tracking target that is included in the prohibited area. has a configuration in which no new identification number is assigned.

The passage processing unit of the passing person measuring device is configured to change at least one of the shape, size, and position of the prohibited area every time Δt elapses according to a predetermined rule. It's okay.

In the above-mentioned passing person measuring device, the area image generated at time t is, for example, a thermal image generated at time t, or a difference image between two thermal images generated at time t-Δt and time t. .

According to the present invention, it is possible to provide a passing person measuring device that is less likely to cause erroneous measurements than conventional ones.

FIG. 1 is a block diagram of a passing person measuring device according to a first embodiment of the present invention. FIG. 3 is a diagram showing an area image generated by the passer-people measuring device according to the first embodiment. FIG. 3 is a diagram related to recognition of a tracking target and assignment of an identification number, which is executed by the control unit of the number-of-passers-measuring device according to the first embodiment. FIG. 3 is a diagram related to position prediction of a tracking target executed by the control unit of the passing person measuring device according to the first embodiment. FIG. 3 is a diagram related to a passing process executed by a control unit of the passing person measuring device according to the first embodiment. FIG. 3 is a diagram showing details of an area image generated when a person approaches the outer edge of a detection area in the passing person measuring device according to the first embodiment. FIG. 4 is a diagram related to a problem that may occur when a prohibited area is not set. 8 is a diagram showing that the problem shown in FIG. 7 can be solved by setting a prohibited area. FIG. FIG. 2 is an operation flow diagram of the passing person measuring device according to the first embodiment. FIG. 2 is a block diagram of a passing person measuring device according to a second embodiment of the present invention.

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of a passing person measuring device according to the present invention will be described with reference to the accompanying drawings.

[First embodiment]
1 shows a passing number counting device 10A according to a first embodiment of the present invention. The passing number counting device 10A is for counting the number of pedestrians W1 passing in a direction M1 and the number of pedestrians W2 passing in a direction M2 through a detection area A provided on a floor 200 near an entrance of a commercial facility, and as shown in the figure, includes a thermal image generating unit 11A, a memory unit 12, a control unit 20A, a setting unit 13, an external interface unit 14, an audio output unit 15, a light emitting unit 16, and a housing that accommodates these. The passing number counting device 10A is attached to a ceiling 100 so that the thermal image generating unit 11A faces the detection area A.

In this example, the mounting height H is 4 m. Furthermore, the detection area A of this embodiment is a rectangle with a length of 4 m on a side perpendicular to the assumed passing directions M1 and M2 and a length of 3 m on a side parallel to the assumed passing directions M1 and M2. When the mounting height H changes, the size of the detection area A also changes accordingly.

The thermal image generation unit 11A is composed of a two-dimensional thermal image sensor. Thermal image generation unit 11A detects far infrared rays emitted from detection area A (floor 200) or pedestrians W1 and W2 in the area A, and generates a 60 pixel x 45 pixel thermal image ( More precisely, it generates data regarding thermal images). Generation of a thermal image is performed every 1/10 seconds. The generated thermal images are sequentially sent to the storage unit 12.

The storage unit 12 is composed of volatile or nonvolatile memory. The storage unit 12 stores at least the latest thermal image and the previous thermal image among the thermal images sent every 1/10 seconds from the thermal image generating unit 11A. In other words, the storage unit 12 stores the thermal image generated last (hereinafter referred to as "second thermal image") and the thermal image generated 1/10 seconds before (hereinafter referred to as "first thermal image"). ) are stored.

The control unit 20A is composed of a microprocessor (MPU) and a computer program executed on the processor. In this embodiment, by executing the computer program, the difference image generation section 21, the tracking target recognition section 22, the identification number assignment section 23, the occurrence position storage section 24, the position prediction section 25, and the passage processing section 26 are installed in the control section 20A. It is formed.

The difference image generation unit 21 generates a difference image P (more precisely, data regarding the difference image P) based on the first thermal image and the second thermal image stored in the storage unit 12. For example, the difference image generation unit 21 subtracts the temperature indicated by the pixel at coordinates (0,0) constituting the first thermal image from the temperature indicated by the pixel at coordinates (0,0) constituting the second thermal image. , the thermal change that should be exhibited by the pixel at coordinates (0, 0) constituting the difference image P is determined.

As shown in FIG. 2, the difference image P is composed of two divided regions of the same size (a first divided region P1 and a second divided region P2) adjacent to each other in the passing directions M1 and M2, and has a total of 60 pixels × It has a size of 45 pixels. Note that the hatched F1 portion is the outer edge region of the first divided region P1, and the hatched F2 portion different from F1 is the outer edge region of the second divided region P2. In this embodiment, among the four sides of the first divided region P1 having a rectangular shape, the outer edge region F1 is a region that is in contact with three sides that are not in contact with the second divided region P2. Similarly, in this embodiment, the outer edge region F2 is a region that touches three sides that do not touch the first divided region P1 among the four sides that constitute the second divided region P2 having a rectangular shape.

In this embodiment, this difference image P is used as an "area image". It can be said that the area image P shows what kind of thermal change occurred in the detection area A during 1/10 seconds.

The tracking target recognition unit 22 recognizes a tracking target corresponding to a person within the detection area A based on the area image P generated every 1/10 seconds. The tracking target recognition unit 22 may perform appropriate noise removal processing on the area image P and then recognize the tracking target.

The identification number assigning unit 23 assigns an identification number to the tracking target recognized by the tracking target recognizing unit 22. When the number of recognized tracking targets is plural, the identification number assigning unit 23 assigns an identification number to each of them.

With reference to FIG. 3, the recognition of a tracking target performed by the tracking target recognition unit 22 and the assignment of an identification number performed by the identification number assigning unit 23 will be specifically described.

The area image _Pt0 generated at time t0 (see FIG. 3A) does not include any pixels indicating that a thermal change has occurred. Therefore, the tracking target recognition unit 22 determines that a group of pixels corresponding to a person in the detection area A (hereinafter referred to as a "cluster"), that is, anything that should be recognized as a tracking target, does not exist in the area image _Pt0 . Since there is no tracking target, the identification number assignment unit 23 does not assign an identification number.

Area image P _t1 (see FIG. 3(B)) generated at time t1 when Δt (=1/10 s) has elapsed from time t0 includes cluster TG1 located in first divided region P1. Therefore, the tracking target recognition unit 22 recognizes the cluster TG1 as a tracking target. Then, the identification number assigning unit 23 assigns an identification number "01" to the cluster TG1, and stores the occurrence location by associating "P1", which is the occurrence location of the cluster TG1, with "01", which is the identification number of the cluster TG1. 24.

The area image P _t2 (see FIG. 3C) generated at time t2 includes a cluster TG1 located in the first divided region P1 and a cluster TG2 located in the second divided region P2. Therefore, the tracking target recognition unit 22 recognizes cluster TG1 and cluster TG2 as tracking targets.

When assigning identification numbers to cluster TG1 and cluster TG2, identification number assigning unit 23 refers to the positions of all clusters (in this specific example, only cluster TG1) included in area image P _t1 generated at time t1. do. Then, the identification number assigning unit 23
・Distance between cluster TG1 at time t1 and cluster TG1 at time t2 ・Distance between cluster TG1 at time t1 and cluster TG2 at time t2 is compared, and the one closest to cluster TG1 at time t1 is time t2 Since the cluster TG1 is at the time t2, the identification number assigned to the cluster TG1 at the time t1, that is, the identification number "01" is assigned to the cluster TG1 at the time t2. On the other hand, the identification number assigning unit 23 assigns a new identification number "02" to cluster TG2, and associates "P2", which is the location of occurrence of cluster TG2, with "02", which is the identification number of cluster TG2. It is stored in the position storage section 24.

Area image P _t3 (see FIG. 3D) generated at time t3 includes cluster TG1 and cluster TG2 located in second divided region P2. Therefore, the tracking target recognition unit 22 recognizes cluster TG1 and cluster TG2 as tracking targets.

When assigning identification numbers to cluster TG1 and cluster TG2, identification number assigning unit 23 calculates the positions of all clusters (cluster TG1 and cluster TG2 in this specific example) included in area image P _t2 generated at time t2. See. Then, the identification number assigning unit 23
- Distance between cluster TG1 at time t2 and cluster TG1 at time t3 - Distance between cluster TG1 at time t2 and cluster TG2 at time t3 - Distance between cluster TG2 at time t2 and cluster TG1 at time t3 Distance - Compare the distance between cluster TG2 at time t2 and cluster TG2 at time t3, and find that cluster TG1 at time t3 is closest to cluster TG1 at time t2, and cluster TG1 at time t2 is closest to cluster TG2 at time t2. Since is cluster TG2 at time t3, the identification number assigned to cluster TG1 at time t2, that is, the identification number "01" is assigned to cluster TG1 at time t3, and the cluster TG2 at time t3 is assigned the identification number "01" at time t2. The identification number assigned to cluster TG2, that is, the identification number "02" is assigned.

In this way, the identification number assigning unit 23 identifies whether the tracking target corresponding to the tracking target included in the current area image, which is the area image P generated at time t, is the area image P generated at time t-Δt. If it is determined that the tracking target is included in the area image, the tracking target included in the current area image is given the same identification number as the tracking target included in the previous area image, and the current If it is determined that the tracking target corresponding to the tracking target included in the area image is not included in the previous area image, a new identification number is given to the tracking target included in the current area image.

The position prediction unit 25 predicts the position at time t+Δt of the tracking target (cluster) included in the current area image, which is the area image P generated at time t.

The passage processing unit 26 determines that the current area image, which is the area image P generated at time t, includes a tracking target whose position at the predicted time t+Δt belongs to an outer edge area of a divided area that is different from the divided area to which the occurrence position belongs. If so, the identification number is deleted from the tracking target, the number of people passing by is counted up by the number of tracking targets, and a prohibited area is set around the position of the tracking target at time t+Δt.

The position prediction performed by the position prediction unit 25 and the passing process performed by the passage processing unit 26 will be specifically described with reference to FIGS. 4 and 5.

At time t3, when an identification number is assigned to each of cluster TG1 and cluster TG2, the position prediction unit 25 calculates the moving distance and moving direction between time t2 and t3 of cluster TG1 assigned identification number "01". That is, the position of cluster TG1 at time t4 is predicted based on the previous movement distance and movement direction. Similarly, the position prediction unit 25 predicts the position of the cluster TG2 at time t4 based on the previous movement distance and movement direction of the cluster TG2 assigned the identification number "02." TG1' and TG2' in FIG. 4(B) indicate predicted positions.

The predicted position TG1' of the cluster TG1 belongs to the outer edge area F2 of the second divided area P2. Furthermore, according to the occurrence position storage unit 24, the occurrence position corresponding to the identification number "01" assigned to cluster TG1 is "P1". Therefore, the passage processing unit 26 deletes the identification number "01" from the cluster TG1 (see FIGS. 4(A) and 5(A)), and also deletes the identification number "01" from the cluster TG1 (see FIG. 4(A) and FIG. 5(A)) and The number of people (pedestrians W1) who passed through the detection area A in the M1 direction (that is, in the M1 direction) is counted up. Thereafter, the passage processing unit 26 sets an elliptical prohibited region NG around the actual cluster TG1 at time t4 (see FIG. 5(B)).

On the other hand, the predicted position TG2' of the cluster TG2 does not belong to either the outer edge area F1 of the first divided area P1 or the outer edge area F2 of the second divided area P2. Therefore, the passage processing unit 26 does not erase the identification number, count up, or set a prohibited area NG for cluster TG2.

Furthermore, even if the predicted position TG2' of the cluster TG2 belongs to the outer edge area F2 of the second divided area P2, the occurrence position corresponding to the identification number "02" given to the cluster TG2 is "P2". , the passage processing unit 26 does not erase the identification number, count up, or set a prohibited area NG for cluster TG2. This is because in such a case, it cannot be said that a person corresponding to cluster TG2 has passed through detection area A.

As described above, the identification number assigning unit 23 assigns an identification number to the tracking target recognized by the tracking target recognizing unit 22. However, the identification number assigning unit 23 assigns an existing identification number to a tracking target belonging to the prohibited area NG, but does not assign a new identification number. The significance of such processing will be explained in detail later.

The external interface unit 14 is connected to appropriate external equipment. The external interface unit 14 transmits the measurement results by the passage processing unit 26, that is, the number of pedestrians W1 passing through the detection area A in the M1 direction and the number of pedestrians W2 passing in the M2 direction, at a predetermined timing and format. can be sent to the relevant external device.

The audio output unit 15 includes a buzzer or a speaker. The audio output unit 15 can notify people in the vicinity of the measurement results by outputting audio corresponding to the measurement results by the passage processing unit 26.

The light emitting section 16 is composed of at least one light emitting diode. The light emitting unit 16 can inform people in the vicinity of the measurement result by lighting or blinking in a manner (color, brightness, etc.) corresponding to the measurement result by the passage processing unit 26.

The setting section 13 is composed of dip switches. The user can perform various settings via the setting section 13. Further, the user can also individually turn on/off the external interface section 14, the audio output section 15, and the light emitting section 16 via the setting section 13.

Next, the significance of setting the prohibited area NG will be explained.

Although clusters are represented by small circles in FIGS. 3 to 5 and FIGS. 7 and 8, which will be referred to later, actual clusters are irregularly spread out that change every moment, like cluster TG1 illustrated in FIG. It has a similar shape. Further, in many cases, the location of the cluster is determined by calculating the center of gravity. Therefore, when a person (TG1) attempting to pass through detection area A in the M1 direction approaches the outer edge of detection area A (outer edge area F2 of second divided area P2), the As shown, the cluster TG1 may appear to remain in the outer edge region F2. The reason for setting the prohibited area NG is to prevent erroneous measurements caused by this.

With reference to FIGS. 7 and 8, it will be specifically explained that erroneous measurements are prevented by setting the prohibited area NG. In this specific example, it is assumed that the cluster TG1 moves in the direction from the first divided area P1 to the second divided area P2, and the cluster TG2 moves to follow it.

If the prohibited area NG is not set, as shown in FIG. 7, the cluster TG1 from which the identification number "01" was deleted at time td is given a new identification number "03" at time te after Δt. Then, at time tf further after Δt,
- The distance between cluster TG1 at time te and cluster TG2 at time tf - The distance between cluster TG2 at time te and cluster TG2 at time tf is compared, and if the former is smaller, the identification number " Another identification number "03" is assigned to the cluster TG2 which had been assigned "02". The identification number "03" is newly assigned to the cluster TG1 belonging to the second divided area P2 at time te, and the occurrence position corresponding to the identification number "03" stored in the occurrence position storage unit 24 is "P2''. Therefore, even if people corresponding to cluster TG2 continue in the M1 direction and exit detection area A, the number of people passing through will not be counted up. In other words, erroneous measurements occur.

On the other hand, when setting a prohibited area NG, as shown in FIG. A new identification number "03" is not assigned to cluster TG1 at time th. Furthermore, at time ti after Δt, cluster TG2, which had been given identification number "02", will not be given another identification number "03", and as a result, the above-mentioned erroneous measurement can be prevented. .

FIG. 9 is an operation flow diagram of the passer-people measuring device 10A according to this embodiment, in which the movements of each part described above are described in chronological order.

In step S1, the storage unit 12 stores the first thermal image generated by the thermal image generation unit 11A.

In step S2, which is executed 1/10 seconds after step S1, the storage section 12 stores the second thermal image generated by the thermal image generation section 11A.

In step S3 executed after step S2, the difference image generation unit 21 generates a difference image (area image) P from the first thermal image and the second thermal image.

In step S4 executed after step S3, the tracking target recognition unit 22 recognizes the tracking target included in the area image P.

In step S5 executed after step S4, the identification number assigning unit 23 assigns an identification number to the recognized tracking target.

In step S6 executed after step S5, the position prediction unit 25 predicts the position of the recognized tracking target 1/10 seconds later.

In step S7, which is executed after step S6, the passage processing unit 26 executes a predetermined passage process based on the predicted position of the tracking target. The passing process includes determining whether the predicted position of the tracking target belongs to the outer edge areas F1, F2, erasing the identification number, counting up the number of people passing through, and setting a prohibited area NG.

The passing person measuring device 10A according to this embodiment repeatedly operates according to this operation flow every 1/10 seconds. In addition, in the flow of the second round, the second thermal image generated and stored in the first round becomes the first thermal image. The same applies to the third and subsequent rounds.

As described above, according to the passing person measuring device 10A according to the present embodiment, a prohibited area NG is set around a tracking target corresponding to a person approaching the outer edge of the detection area A, and a new identification number is assigned. By prohibiting this, it is possible to prevent incorrect measurements from being given to subsequent tracking targets.

[Second example]
FIG. 10 shows a passing person measuring device 10B according to a second embodiment of the present invention. The number-of-passers counting device 10B includes a thermal image generation section 11B instead of the thermal image generation section 11A, a control section 20B instead of the control section 20A, and does not include the storage section 12. Although it is different from the passer-by person measuring device 10A in this respect, it is common to the passer-by person measuring device 10A in other respects.

The thermal image generation unit 11B is capable of generating an image related to a thermal change in the detection area A that occurred during a predetermined time (for example, 1/10 s), that is, an image corresponding to the difference image of the first embodiment. It consists of a two-dimensional thermal image sensor of the type that can be used. In this embodiment, the image generated by the thermal image generation unit 11B is the "area image" P.

The control unit 20B, like the control unit 20A, is composed of a microprocessor and a computer program executed on the processor. In this embodiment, a tracking object recognition section 22, an identification number assignment section 23, an occurrence position storage section 24, a position prediction section 25, and a passage processing section 26 are formed in the control section 20B by executing a computer program.

According to the passer person measuring device 10B according to the second embodiment, the same effects as those of the passer person measuring device 10A according to the first embodiment can be obtained.

[Modified example]
Although the first and second embodiments of the people counting device according to the present invention have been described above, the configuration of the present invention is not limited thereto.

For example, the passing person measuring devices 10A and 10B may measure the number of people passing through a detection area A provided near the entrances and exits of various facilities other than commercial facilities and various vehicles.

Moreover, the mounting location of the passing person measuring devices 10A and 10B may be a wall surface.

Furthermore, the passing person measuring devices 10A and 10B do not need to include all or part of the setting section 13, the external interface section 14, the audio output section 15, and the light emitting section 16.

Further, the number-of-passers counting device 10A does not need to include the difference image generation section 21 (and the storage section 12). In this case, the control unit 20A can use the thermal image itself generated by the thermal image generation unit 11A as the area image P.

Furthermore, the passing person measurement devices 10A and 10B can generate an area image P showing the movement of people in the detection area A instead of the thermal image generation units 11A and 11B, the storage unit 12, and the difference image generation unit 21. An arbitrary area image generation means may be provided. Such means include, for example, a distance image sensor. When a distance image sensor is used, a collection of pixels indicating that the object is higher than the floor 200 becomes a cluster (tracking target).

Further, the detection area A may be composed of two or more divided regions having mutually different shapes.

Further, the outer edge region of each divided region constituting the area image P does not have to be provided so as to touch all sides that do not touch other divided regions. In other words, the outer edge area of each divided area may be provided so as to be in contact with one or more sides that are not in contact with other divided areas.

Furthermore, the following modification examples are possible regarding the prohibited area NG set by the passage processing unit 26 in the situations shown in FIGS. 4 and 5. Note that these modifications can be combined as appropriate.
- Set the shape of the prohibited area NG to be circular or regular polygonal.
- A prohibited area NG is set around cluster TG1 at time t3 instead of around cluster TG1 at time t4.
- A prohibited area NG is set around the predicted position TG1' at time t3.
・Continue setting prohibited areas NG until predetermined conditions are met. For example, the prohibited area NG continues to be set until Δt×3 has elapsed. Alternatively, the prohibited area NG continues to be set until cluster TG1 no longer exists.
- At least one of the shape, size, and position of the prohibited area NG is changed every time Δt elapses according to a predetermined rule. For example, the size of the prohibited area NG is changed depending on the size of the cluster TG1.

10A, 10B Passing person measuring device 11A, 11B Thermal image generation section 12 Storage section 13 Setting section 14 External interface section 15 Audio output section 16 Light emitting section 20A, 20B Control section 21 Difference image generation section 22 Tracking target recognition section 23 Identification number assignment Section 24 Occurrence position storage section 25 Position prediction section 26 Passage processing section 100 Ceiling 200 Floor P Area image P1 First divided region P2 Second divided region TG1, TG2 Tracking target (cluster)
W1, W2 pedestrian

Claims (4)

A passing person measuring device that measures the number of people passing through a predetermined detection area,
a tracking object recognition unit that recognizes a tracking object corresponding to the person based on an area image related to the detection area that is generated every time Δt elapses;
an identification number assigning unit that assigns an identification number to the recognized tracking target;
an occurrence position storage unit that stores to which of a plurality of divided regions constituting the area image the recognized occurrence position of the tracking target belongs in association with an identification number assigned to the tracking target;
a position prediction unit that predicts the position at time t+Δt of a tracking target included in a current area image that is an area image generated at time t;
When the current area image includes a tracking target whose position at the predicted time t+Δt belongs to an outer edge area of a divided area different from the divided area to which the occurrence position belongs, erasing the identification number from the tracking target and , a passage processing unit that counts up the number of people by the number of tracking targets, and further sets a prohibited area around the position of the tracking target at time t or time t+Δt;
Equipped with
When determining that the tracking target corresponding to the tracking target included in the current area image is included in the previous area image that is the area image generated at time t-Δt, the identification number assigning unit The tracking target included in the area image is given the same identification number as the tracking target included in the previous area image, while the tracking target corresponding to the tracking target included in the current area image is assigned. is not included in the previous area image, assigning a new identification number to the tracking target included in the current area image,
Further, in the passing person measuring device, the identification number assigning unit does not assign a new identification number to a tracking target belonging to the prohibited area. 2. The passage processing unit is capable of changing at least one of the shape, size, and position of the prohibited area every time Δt elapses according to a predetermined rule. A device to measure the number of people passing by. 3. The device for measuring the number of people passing by according to claim 1, wherein the area image generated at time t is a thermal image generated at time t. The device for measuring the number of people passing by according to claim 1 or 2, wherein the area image generated at time t is a difference image between two thermal images generated at time t-Δt and time t. .

Images