JP6599644B2 - Motion determination device, motion determination method, and motion determination program - Google Patents

Description

  The present invention relates to a technique for determining whether or not an object captured in a moving image captured by a security camera or the like is abnormal.

  Conventionally, security cameras are installed in various places where an unspecified number of people gather, such as airports, stations, shopping centers, street corners, and the like.

  In addition, it has been studied to process a moving image captured by a security camera and detect a person who has taken a specific action as a suspicious person. For example, a moving image picked up by a camera is processed, and the moving speed, moving direction, head height, etc. of the person being picked up are detected as feature quantities of the person's behavior, and the feature quantities detected here are used. Some of them determine whether or not the person is a suspicious person (see Patent Document 1).

  Note that Patent Document 1 processes a moving image captured by an in-vehicle camera, not a security camera.

JP2013-88870A

  However, the technique described in Patent Document 1 and the like uses a moving speed and a moving direction of a person (hereinafter, referred to as a tracking target person) captured in a moving image, and the behavior of the tracking target person is peculiar. It is the structure which determines whether it is. In order to obtain the movement speed of the tracking target person, the movement amount of the tracking target person in the real space is required. That is, in order to determine whether or not the behavior of the tracking target person is unique, in order to calculate the movement amount of the tracking target person in the real space, the position of the tracking target person in the frame image is set to the position in the real space. Need to be converted to

  Also, the conversion parameter that associates the position on the frame image captured by the security camera with the position in the real space varies depending on the angle of the security camera. Moreover, the angle of the security camera may change after installation due to the influence of external forces such as wind and vibration. Therefore, when the security camera is installed or when the angle of the security camera changes, the conversion parameter setting work is performed, which requires manpower and cost for operation management.

  As is well known, even if the movement amount of the tracking target person in the real space is the same, the movement amount of the tracking target person on the frame image changes according to the positional relationship between the camera and the tracking target person. . That is, the movement amount of the tracking target person in the real space cannot be directly calculated from the movement amount of the tracking target person on the frame image.

  An object of the present invention is to provide a technique capable of easily determining whether or not an object captured in a moving image is unusual and reducing manpower and cost for operation management. is there.

  In order to achieve the above object, the motion determination apparatus of the present invention is configured as follows.

  A moving image obtained by imaging the target area is input to the moving image input unit. The moving image input to the moving image input unit is, for example, a moving image of a target area captured by a security camera.

  The moving direction detection unit processes a frame image applied to the moving image input to the moving image input unit, and detects a moving direction on the frame image for each of a plurality of detection areas defined for the object being picked up. . The detection area is set at a position overlapping the tracking target person or around the tracking target person. For example, the target area is set to a rectangular area surrounding the whole body of the tracking target person on the frame image or a rectangular area surrounding the upper body. Then, the detection area is set to a divided area obtained by dividing the target area into a matrix. Thereby, a plurality of detection areas are set on one frame image.

The moving direction θ (0 ≦ θ <2π) of the detection area between two frame images with different imaging timings is the position (x1, y1) of the detection area on one frame image and the other frame image. Using the position (x2, y2) of this detection area,
θ = arg tan ((y2-y1) / (x2-x1))
Can be calculated.

The variation degree calculation unit calculates the degree of variation in the movement direction detected by the movement direction detection unit for a plurality of detection regions defined for the object. The degree of variation calculated here may be the L0 norm of the histogram of the movement direction detected for each detection area, or may be the variance δ ² or the standard deviation δ of the movement direction detected for each detection area.

  The determination unit determines whether or not the movement of the object is unique using the variation degree calculated by the variation degree calculation unit. Comparing a person with a peculiar movement with a person without a peculiar movement, a person with a peculiar movement moves each part of the body compared to a person who does not peculiar movement There are variations in the direction. In addition, people who have a specific movement here are people who are running, people who are staggered (people who are drunk and stray, people who are staggered due to poor physical condition, etc.) Etc.).

  As described above, in this configuration, the movement of the object is unique without using the position of the object captured in the frame image in the real space (without detecting the moving speed of the object in the real space). It can be determined whether or not there is. Therefore, it is possible to easily determine whether or not the motion of the object captured in the moving image is unique, and to reduce manpower and cost for operation management.

  In addition, for each of a plurality of detection areas defined for the object, a movement amount detection unit that detects a movement amount on the frame image, and a movement detected by the movement amount detection unit for the plurality of detection areas defined for the object A movement specific value calculation unit that calculates a movement specific value of the object from the amount, and the determination unit determines whether the movement of the object is singular, the variation degree calculated by the variation degree calculation unit, and the movement specific value You may make it the structure which determines using the movement specific value which the calculation part calculated.

  In this case, it is possible to determine whether or not the movement of the object is unique in consideration of the amount of movement of the object on the frame image. As a result, it is possible to accurately determine the person who is running as a person who is making a specific movement.

The amount of movement of the detection region between two frame images with different imaging timings (the amount of movement on the frame image) is the position (x1, y1) of the detection region on one frame image and the other frame image. Using the position (x2, y2) of this detection area,
Movement amount = ((x2−x1) ² + (y2−y1) ² ) ^1/2
Can be calculated.

  In addition, the movement specific value calculated by the movement specific value calculation unit may be the maximum value of the movement amount on the frame image detected for each detection region, may be an average value, or may be a median value.

  Further, the determination unit may be configured to determine whether or not the movement of the object is unique, for example, by comparing a value obtained by multiplying the degree of variation and the movement specific value with a determination threshold value. In this case, the determination threshold is preferably determined according to the moving direction of the object on the frame image.

  According to the present invention, it is possible to easily determine whether or not the motion of an object captured in a moving image is peculiar, and to reduce manpower and cost for operation management.

It is a block diagram which shows the structure of the principal part of a motion determination apparatus. It is a figure which shows the function structure of an image process part. It is a figure explaining the moving direction on a frame image. It is a figure explaining the moving direction of the detection area of the person who is walking. It is a figure explaining the moving direction of the detection area of the person who is running. It is a figure which shows the histogram of the moving direction of the detection area of the person who is walking. It is a figure which shows the histogram of the moving direction of the detection area | region of the person who is running. It is a flowchart which shows operation | movement of the image process part of a motion determination apparatus.

  Hereinafter, a motion determination apparatus according to an embodiment of the present invention will be described.

  FIG. 1 is a block diagram illustrating a configuration of a main part of a motion determination device according to this example. The motion determination device 1 includes a moving image input unit 2, an image processing unit 3, and an output unit 4. The motion determination device 1 according to this example processes a moving image captured by the imaging device 10 such as a security camera, and the motion of a person (corresponding to an object referred to in the present invention) captured in the moving image is unique. And the determination result is output to a higher-level device or the like.

  In addition, the person who is moving here is a person who is running, a person who is swaying (a person who is drunk or swaying, a person who is swaying due to poor physical condition, etc.) Who are fighting).

  An imaging device 10 is connected to the moving image input unit 2. The imaging device 10 is a video camera that captures moving images. The imaging device 10 is a security camera installed at a place (an airport, a station, a shopping center, a street corner, etc.) where, for example, an unspecified number of people gather. The imaging device 10 inputs the captured moving image to the moving image input unit 2. That is, a frame image related to a moving image captured by the imaging device 10 is input to the moving image input unit 2 in time series. The frame rate of the imaging device 10 may be about 5 to 30 frames / sec. The imaging area of the imaging device 10 corresponds to the target area referred to in the present invention.

  The moving image input unit 2 is connected to a playback device that plays back a moving image file recorded on a medium such as a USB memory, an SD memory card, a DVD, or an HD, and a playback image ( The playback image of the moving image file) may be input. For example, a moving image file relating to a moving image of an imaging area captured by the imaging device 10 is recorded on the medium. Moreover, the structure which has the function to reproduce | regenerate the input moving image file may be sufficient as the motion determination apparatus 1 concerning this example.

  The image processing unit 3 processes a frame image applied to the moving image input to the moving image input unit 2 to track the movement of the person being imaged, and whether or not the movement of the tracked person is unique The process etc. which determine are performed. Details of these processes will be described later.

  The image processing unit 3 includes a computer that executes the motion determination method according to the present invention. The motion determination program according to the present invention is installed in a computer included in the image processing unit 3.

  Note that the image processing unit 3 may process all frame images related to the moving image input to the moving image input unit 2 as processing target frame images, or the moving image input to the moving image input unit 2 Of these frame images, frame images extracted at a predetermined interval may be processed as processing target frame images (frame images not extracted here are not subjected to processing described later). The interval for extracting the processing target frame image may be determined according to the frame rate of the moving image input to the moving image input unit 2, and may be set at a time interval or a frame interval. May be.

  The output unit 4 outputs the determination result in the image processing unit 3. For example, the output unit 4 notifies the host device of the presence or absence of a person whose movement is determined to be unique by the image processing unit 3. For example, when the image processing unit 3 is notified that there is a person whose motion is unique, the host device is configured to notify that a person whose motion is unique has been detected. This notification may be performed by turning on an indicator lamp, sending a voice message from a speaker, or the like. The output unit 4 may be configured to display a moving image input to the moving image input unit 2 on a connected display device (not shown). In this case, the motion determination apparatus 1 may be configured to output from the output unit 4 a video signal that highlights a person whose motion detected by the image processing unit 3 is unique.

  Next, the functional configuration of the image processing unit 3 will be described. FIG. 2 is a block diagram illustrating a functional configuration of the image processing unit. The image processing unit 3 includes an object tracking function unit 31, a detection area setting function unit 32, a movement direction detection function unit 33, a variation degree calculation function unit 34, a movement amount detection function unit 35, and a movement specific value calculation function. Unit 36, determination value calculation function unit 37, determination threshold value determination function unit 38, and determination function unit 39.

  The object tracking function unit 31 detects a person captured in the processing target frame image as a tracking target person for each processing target frame image extracted from the frame image of the moving image captured by the imaging device 10. As described above, the processing target frame image may be all frame images related to the moving image input to the moving image input unit 2 or may be determined in advance from the moving image input to the moving image input unit 2. Frame images extracted at intervals may be used. The object tracking function unit 31 detects a tracking target person captured in the processing target frame image by, for example, a known background difference method.

  Further, the object tracking function unit 31 performs an identification process for associating a captured tracking target person between temporally continuous processing target frame images. Further, the object tracking function unit 31 creates an object map for tracking the movement of the tracking target person. The object map includes, for each tracking target person, a processing target frame image in which the tracking target person is detected, and a position of the tracking target person on the processing target frame image (two-dimensional coordinates (x, y) on the processing target frame image). ) Are registered in association with each other. The position of the tracking target person registered in the object map may be, for example, the position of the top of the tracking target person on the processing target frame image (a position that is less susceptible to occlusion). From the object map created by the object tracking function unit 31, the movement direction and movement amount of the tracking target person on the processing target frame image can be obtained.

  The object tracking function unit 31 does not need to calculate the position of the tracking target person in the real space for the tracking target person captured in the processing target frame image.

  The detection region setting function unit 32 sets a plurality of detection regions for detecting the moving direction for each tracking target person captured in the processing target frame image detected by the object tracking function unit 31. . The detection area setting function unit 32 sets a target area (rectangular area) on the processing target frame image. This target area is set for each tracking target person. For example, the target region may be a region surrounding the whole body of the tracking target person on the processing target frame image, or may be a region surrounding the upper body. A method for setting the target area is predetermined. The detection area setting function unit 32 divides the target area into a matrix and sets each divided area as a detection area. The number of divisions of the target region (the number of divisions in the vertical direction and the horizontal direction) may be determined in advance. Therefore, the detection area is set on the tracking target person on the processing target frame image or in the vicinity thereof. The total number of detection areas set for the tracking target person may be a number that can appropriately calculate the degree of variation in the moving direction, which will be described later.

  It should be noted that the movement direction of the tracking target person obtained from the object map created by the object tracking function unit 31 and the movement direction of each detection area set for the tracking target person are not always the same.

  The movement direction detection function unit 33 detects the movement direction of each detection region set for the tracking target person for each tracking target person. The moving direction of the detection area is the direction in which the position of the detection area has changed between two temporally continuous processing target frame images (direction on the processing target frame image). For example, as shown in FIG. 3, the moving direction is an angle θ (0 ≦ θ <360 °) set clockwise with the rightward horizontal direction on the frame image being 0 °. The reference of the angle θ (direction to be 0 °) is not limited to the right horizontal direction on the frame image, and can be arbitrarily set. The angle θ is relative to the reference direction. You may set it counterclockwise.

The movement direction detection function unit 33 detects each detection area movement direction set for the tracking target person by a known optical flow. The moving direction θ (0 ≦ θ <2π) of the detection area on the frame image is determined based on the position (x1, y1) of the detection area on the previous processing target frame image and the current processing target frame image. Using the position (x2, y2) of the detection area,
θ = arg tan ((y2-y1) / (x2-x1))
Can be calculated.

  For example, as shown in FIG. 4, when the tracking target person is a walking person, the movement directions are substantially the same in many detection regions set for the tracking target person. Therefore, when the tracking target person is a walking person, the variation in the moving direction detected for the plurality of detection regions is relatively small.

  On the other hand, as shown in FIG. 5, when the tracking target person is a running person, the movement direction of the detection area set for the tracking target person is various directions. This is because the person to be tracked moves each part of the body relatively intensely. Therefore, when the tracking target person is a running person, the variation in the moving direction detected for the plurality of detection areas is relatively large.

  In addition to running people, people who are flirting (people who are drunk and flirting, people who are flirting due to poor physical condition, etc.), people who are fighting (people who are fighting), etc. Since the movement is relatively intense, the variation in the movement direction detected for the plurality of detection areas is relatively large. Moreover, the arrows shown in FIGS. 4 and 5 indicate the moving directions detected for some detection areas for convenience, and are not captured in the processing target frame image.

  As is apparent from the above description, the movement direction detection function unit 33 does not need to calculate the position of the tracking target person in the real space when detecting the movement direction of each detection region.

  The variation degree calculation function unit 34 calculates, for each tracking target person, the degree of variation in the movement direction of each detection region set for the tracking target person in the same processing target frame image. In this example, the L0 norm of a 36-dimensional histogram obtained by dividing the movement direction of each detection area of the tracking target person at 10 ° intervals is calculated as the degree of variation. The L0 norm is a non-zero dimension number. That is, in this example, the calculated variation degree is an integer in the range of 1 ≦ variation degree ≦ 36.

  FIG. 6 is an example of a histogram created for a walking person shown in FIG. FIG. 7 is an example of a histogram created for a running person shown in FIG. As shown in FIG. 6, the person walking is concentrated in the direction of movement of the detection area centering around 180 °. On the other hand, as shown in FIG. 7, the running person has a different movement direction of the detection region as compared to the walking person shown in FIG. The L0 norm of the histogram shown in FIG. 6 is 14, and the L0 norm of the histogram shown in FIG.

  In this example, the moving direction of the detection region is divided into 36 dimensions, but may be divided into 24 dimensions (15 ° intervals), 72 dimensions (5 ° intervals), and the like. Further, the total number of detection regions set for the tracking target person may be about several times (1 to 3 times) the number of dimensions. The number of dimensions and the total number of detection areas are preferably set so that the difference in the degree of variation in the movement direction of each detection area is large between a walking person and a running person.

  If the number of dimensions to be divided is too small, the range of one dimension becomes wide, and the movement direction of each detection region is easily concentrated on a specific dimension. On the other hand, if the number of dimensions to be divided is increased too much, the range of one dimension becomes small, and therefore the movement direction of each detection region is likely to be distributed over many dimensions. Therefore, even if the number of dimensions to be divided is too small or too large, the difference in the degree of variation in the movement direction of each detection region between the walking person and the running person becomes small.

  The movement amount detection function unit 35 detects the movement amount of each detection area set for the tracking target person for each tracking target person. The amount of movement of the detection region is the amount of change in the position of the detection region between two temporally continuous processing target frame images (the size on the processing target frame image).

The movement direction detection function unit 33 detects the movement amount of the corresponding detection area in the previous process target frame image and the current process target frame image. The amount of movement of the detection area on the frame image is determined based on the position (x1, y1) of the detection area on the previous processing target frame image and the position (x2, y2) of the detection area on the current processing target frame image. ) And
Movement amount = ((x2−x1) ² + (y2−y1) ² ) ^1/2
Can be calculated.

  In this example, since the movement direction detection function unit 33 described above performs an optical flow when detecting the movement direction of each detection region, the movement amount detection function unit 35 is performed by the movement direction detection function unit 33. The amount of movement of each detection region may be detected using the result of the optical flow. That is, the movement amount detection function unit 35 does not need to perform an optical flow.

  However, the movement amount detection function unit 35 may perform an optical flow. In this case, the movement direction detection function unit 33 may detect the movement direction of each detection region using the result of the optical flow performed by the movement amount detection function unit 35. That is, the optical flow may be performed by one of the movement direction detection function unit 33 or the movement amount detection function unit 35, and the other may use the result.

  As is apparent from the above description, the movement amount detection function unit 35 does not need to calculate the position of the tracking target person in the real space when detecting the movement amount of each detection region.

  The movement specific value calculation function unit 36 calculates a movement specific value based on the movement amount detected by the movement amount detection function unit 35 for each detection region set for the tracking target person. In this example, the movement specific value is an average value of the movement amount of each detection region.

  The movement specific value may be the maximum value or the median value of the movement amount of each detection region.

The determination value calculation function unit 37 calculates a determination value using the variation degree calculated by the variation degree calculation function unit 34 and the movement specific value calculated by the movement specific value calculation function unit 36. In this example, the determination value is a value obtained by multiplying the variation degree calculated by the variation degree calculation function unit 34 and the movement specific value calculated by the movement specific value calculation function unit 36. That is,
Judgment value = (movement specific value) x (variation)
It is.

  Note that the determination value may be a value calculated by a calculation formula other than the above.

The determination threshold value determination function unit 38 determines based on the moving direction of the tracking target person on the frame image (not the moving direction of each detection area described above). The moving direction of the tracking target person on the frame image is obtained from the object map created by the object tracking function unit 31 as described above. In this example,
Determination threshold = A × | cos α | + B
The value calculated by is determined. However, A and B are predetermined constants, and α is the movement direction of the tracking target person on the frame image. The moving direction α of the tracking target person on the frame image is also an angle α (0 ≦ α <360 °) set clockwise with the rightward horizontal direction on the frame image shown in FIG. 3 being 0 °.

  When a tracking target person who moved in the horizontal direction (horizontal direction shown in FIG. 3) on the frame image is compared with a tracking target person who moved in the vertical direction on the frame image, these tracking target persons When the amount of movement is the same, the amount of movement is greater for the tracking target person who has moved in the horizontal direction on the frame image. This is because the tracking target person who moved in the vertical direction on the frame image has a larger moving component in the vanishing point direction. Therefore, the determination threshold value determined by the determination threshold value determination function unit 38 is as the moving direction of the tracking target person on the frame image approaches the vertical direction (as approaching 90 ° or 270 ° shown in FIG. 3). I try to make it smaller.

  Note that the determination threshold value determination function unit 38 may be configured to calculate the determination threshold value using another calculation formula as long as a condition that decreases as the moving direction of the tracking target person on the frame image becomes the vertical direction is satisfied.

  The determination function unit 39 compares the determination value calculated by the determination value calculation function unit 37 with the determination threshold determined by the determination threshold determination function unit 38. If the determination value is less than the determination threshold, the movement of the tracking target person Is determined to be non-singular, and if the determination value is equal to or greater than the determination threshold, it is determined that the movement of the person being tracked is peculiar.

  FIG. 8 is a flowchart showing the operation of the image processing unit of the motion determination apparatus. The image processing unit 3 determines the current processing target frame image from the frame images related to the moving image input to the moving image input unit 2 (s1).

  The image processing unit 3 processes the current processing target frame image determined in s1, and performs tracking processing for the tracking target person (s2). In s2, the object map described above is created. The object tracking function unit 31 performs the process related to s2.

  The image processing unit 3 sets a detection area for each tracking target person (s3). In s3, the detection area setting function unit 32 sets a detection area for detecting a movement vector by optical flow. In this example, the detection area setting function unit 32 sets, for each tracking target person captured in the processing target frame image, a rectangular area surrounding the upper body as the target area for the tracking target person. The detection area setting function unit 32 divides the set target area into a matrix and sets each divided area as a detection area.

  For each tracking target person, the image processing unit 3 calculates the movement specific value of the tracking target person (s4) and calculates the degree of variation in the movement direction of the tracking target person (s5). The processes according to s4 and s5 may be executed in order or may be performed in parallel. The process related to s4 is performed by the movement amount detection function unit 35 and the movement specific value calculation function unit. Further, the processing related to s5 is performed by the movement direction detection function unit 33 and the variation degree calculation function unit 34.

  The process related to the optical flow for detecting the movement vector of each detection area is performed prior to the processes related to s4 and s5. One of the movement amount detection function unit 35 and the movement direction detection function unit 33 performs the process related to the optical flow.

  In s4, the movement amount of each detection area set for the tracking target person is detected for each tracking target person, and the average value is calculated as the movement specific value of the tracking target person. In s5, the movement direction of each detection region set for the tracking target person is detected for each tracking target person, and the L0 norm of the histogram shown in FIGS. Calculated as the degree of variation in the moving direction.

  The image processing unit 3 calculates a determination value for each tracking target person using the movement specific value calculated in s4 and the variation degree calculated in s5 (s6). The determination value calculated in s6 is a value obtained by multiplying the movement specific value calculated in s4 and the variation degree calculated in s5. The determination value calculation function unit 37 performs the process related to s6.

  The image processing unit 3 determines a determination threshold for each tracking target person for each tracking target person (s7). As described above, the determination threshold is determined according to the movement direction of the tracking target person on the frame image. The process related to s7 may be performed in parallel with the process related to s2 to s6 described above, or may be executed before starting any of the processes of s2 to s6. The determination threshold value determination function unit 38 performs the process related to s7.

  For each tracking target person, the image processing unit 3 compares the determination value calculated in s6 for the tracking target person with the determination threshold calculated in s7, and determines whether the behavior of the tracking target person is unique. A determination process for determining is performed (s8). The determination function unit 39 performs the process related to s8. In s8, if the determination value calculated in s6 for each tracking target person is less than the determination threshold value calculated in s7, it is determined that the action is not unique and the determination threshold value calculated in s7 is greater than or equal to the determination threshold value. For example, it is determined that the behavior is unique.

  If the image processing unit 3 determines in s8 that the behavior is unique for at least one tracking target person, the image processing unit 3 outputs a determination result indicating that there is a tracking target person whose behavior is unique (s9, s11). ). On the other hand, when the image processing unit 3 determines in s8 that the behavior is not unique for all the tracking target persons, the image processing unit 3 outputs a determination result indicating that there is no tracking target person whose behavior is unique (s9, s10). ).

  As described above, the motion determination apparatus 1 according to this example does not calculate the position in the real space for the tracking target person captured in the frame image related to the input moving image, and the behavior of the tracking target person is calculated. Can be determined whether or not. Therefore, it is possible to easily determine whether or not the movement of the tracking target person (object) captured in the input moving image is peculiar, and to reduce the manpower and cost for operation management.

In the above example, the variation degree of the movement direction calculated in s5 is the L0 norm of the movement direction of each detection area, but may be a variance δ ² or a standard deviation δ of the movement direction of each detection area.

  The determination value calculated in s6 may be the degree of variation in the movement direction of each detection region calculated in s5. This eliminates the need for detecting the amount of movement for each detection region, so that the processing load can be reduced and the cost of the main body of the motion determination apparatus 1 can be reduced.

  Further, the determination threshold value calculated in s7 may be a preset fixed value. In this way, the configuration for calculating the determination threshold value for each person to be tracked becomes unnecessary, so that the processing load can be reduced and the cost of the main body of the motion determination device 1 can be reduced.

DESCRIPTION OF SYMBOLS 1 ... Motion determination apparatus 2 ... Moving image input part 3 ... Image processing part 4 ... Output part 10 ... Imaging device 31 ... Object tracking function part 32 ... Detection area setting function part 33 ... Moving direction detection function part 34 ... Variability calculation function Unit 35 ... Movement amount detection function unit 36 ... Movement specific value calculation function unit 37 ... Determination value calculation function unit 38 ... Determination threshold value determination function unit 39 ... Determination function unit

Claims (9)

A moving image input unit for inputting a moving image obtained by imaging the target area;
A moving direction detection unit that processes a frame image related to the moving image input to the moving image input unit and detects a moving direction on the frame image for each of a plurality of detection regions defined for the object being imaged; ,
A variation degree calculation unit that calculates a variation degree of the movement direction detected by the movement direction detection unit for the plurality of detection regions defined for the object;
Whether the movement of the object is singular, Bei give a, a determination unit by using the variation degree of the variation degree calculating unit is calculated,
The variation degree calculating unit calculates a L0 norm of a moving direction histogram detected by the moving direction detection unit as the degree of variation for the plurality of detection regions defined for the object . A movement amount detection unit that detects a movement amount on a frame image for each of the plurality of detection regions defined for the object;
A movement specific value calculation unit that calculates a movement specific value of the object from the movement amount detected by the movement amount detection unit for the plurality of detection areas defined for the object;
The determination unit determines whether or not the movement of the object is unique using the variation degree calculated by the variation degree calculation unit and the movement specific value calculated by the movement specific value calculation unit. Item 2. The motion determination device according to Item 1.   The movement according to claim 2, wherein the determination unit determines whether or not the movement of the object is unique by comparing a determination value obtained by multiplying the degree of variation and the movement specific value with a determination threshold. Judgment device.   The motion determination apparatus according to claim 3, further comprising a determination threshold value determination unit that determines the determination threshold value according to a moving direction of the object on a frame image. A moving image input unit for inputting a moving image obtained by imaging the target area;
A moving direction detection unit that processes a frame image related to the moving image input to the moving image input unit and detects a moving direction on the frame image for each of a plurality of detection regions defined for the object being imaged; ,
A variation degree calculation unit that calculates a variation degree of the movement direction detected by the movement direction detection unit for the plurality of detection regions defined for the object;
A movement amount detection unit that detects a movement amount on a frame image for each of the plurality of detection regions defined for the object;
A movement specific value calculation unit that calculates a movement specific value of the object from the movement amount detected by the movement amount detection unit for the plurality of detection areas defined for the object;
A determination value obtained by multiplying the variation degree calculated by the variation degree calculation unit by the movement specific value calculated by the movement specific value calculation unit, and a determination threshold value, whether or not the movement of the object is peculiar. A determination unit for determination by comparison;
A determination threshold value determining unit configured to determine the determination threshold value according to a moving direction of the object on a frame image; A frame image related to a moving image obtained by imaging the target area input to the moving image input unit is processed, and a moving direction on the frame image is detected for each of a plurality of detection areas defined for the object being imaged. A moving direction detection step;
A variation degree calculating step for calculating a variation degree of the movement direction detected in the movement direction detection step for the plurality of detection regions defined for the object;
A determination step of determining whether or not the movement of the object is peculiar using the variation degree calculated in the variation degree calculation step ,
The variation degree calculating step is a step of calculating the L0 norm of the moving direction histogram detected in the moving direction detecting step as the degree of variation for the plurality of detection regions defined for the object. Method. A frame image related to a moving image obtained by imaging the target area input to the moving image input unit is processed, and a moving direction on the frame image is detected for each of a plurality of detection areas defined for the object being imaged. A moving direction detection step;
A variation degree calculating step for calculating a variation degree of the movement direction detected in the movement direction detection step for the plurality of detection regions defined for the object;
A variation degree calculating step for calculating a variation degree of the movement direction detected in the movement direction detection step for the plurality of detection regions defined for the object;
A movement amount detecting step for detecting a movement amount on a frame image for each of the plurality of detection regions defined for the object;
A movement specific value calculation step for calculating a movement specific value of the object from the movement amount detected in the movement amount detection step for the plurality of detection regions defined for the object;
A determination value obtained by multiplying the variation degree calculated in the variation degree calculation step by the movement specific value calculated in the movement specific value calculation step, and a determination threshold value, whether the movement of the object is peculiar. A determination step for determining by comparison;
And a determination threshold value determining step for determining the determination threshold value according to the moving direction of the object on the frame image. A frame image related to a moving image obtained by imaging the target area input to the moving image input unit is processed, and a moving direction on the frame image is detected for each of a plurality of detection areas defined for the object being imaged. A moving direction detection step;
A variation degree calculating step for calculating a variation degree of the movement direction detected in the movement direction detection step for the plurality of detection regions defined for the object;
Determining whether or not the movement of the object is peculiar using the variation degree calculated in the variation degree calculation step, and causing the computer to execute
The variation degree calculating step is a step of calculating the L0 norm of the moving direction histogram detected in the moving direction detecting step as the degree of variation for the plurality of detection regions defined for the object. program. A frame image related to a moving image obtained by imaging the target area, which is input to the moving image input unit, is processed, and a moving direction on the frame image is detected for each of a plurality of detection areas determined for the imaged object. A moving direction detection step;
A variation degree calculating step for calculating a variation degree of the movement direction detected in the movement direction detection step for the plurality of detection regions defined for the object;
A variation degree calculating step for calculating a variation degree of the movement direction detected in the movement direction detection step for the plurality of detection regions defined for the object;
A movement amount detecting step for detecting a movement amount on a frame image for each of the plurality of detection areas defined for the object;
A movement specific value calculation step for calculating a movement specific value of the object from the movement amount detected in the movement amount detection step for the plurality of detection regions defined for the object;
A determination value obtained by multiplying the variation degree calculated in the variation degree calculation step by the movement specific value calculated in the movement specific value calculation step, and a determination threshold value, whether the movement of the object is peculiar. A determination step for determining by comparison;
A motion determination program for causing a computer to execute a determination threshold value determining step of determining the determination threshold value according to a moving direction of the object on a frame image.