JP2003044859A - Device for tracing movement and method for tracing person - Google Patents

Abstract

PROBLEM TO BE SOLVED: To quickly and surely enable a tracing device to issue notification in the case of judging that an object intruding into a detection area seems to be a person. SOLUTION: An object extracting part 105 extracts an object with respect to an image of the detection area photographed by an imaging unit 101, and an object deciding part 108 decides whether or not the object is a person on the basis of characteristic quantity, such as an area ration, aspect ratio, moving quantity, moving speed, center of gravity and height calculated by a characteristic quantity calculating part 107. As a result, the object judging part 108 continues tracing processing without issuing notification even through the object decided to be a person is outside the detection are and also adds the number of tracing processing times. In the case that the extracted object is within the detection area, that the object is decided to be a person and that the object is the same as the object previously subjected to tracing processing, notification is issued when the number of tracing processing times including tracing processing outside the detection area becomes equal to or more than the prescribed number of times.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a device for photographing and monitoring a specific place, and more specifically, it processes an image obtained from an image pickup device to extract a moving object for tracking processing.
The present invention relates to a motion tracking device and a person tracking method for notifying when a person enters a predetermined area.

[0002]

2. Description of the Related Art By monitoring a specific place, extracting a person who has invaded the place, and performing tracking processing, a warning device such as a buzzer is emitted, or a monitoring device or person tracking for notifying an upper system of an alarm or the like. The device is known.

FIG. 9 is a block diagram showing a configuration example of a conventional person tracking device. In this configuration, the input unit 902
Is an input image storage unit 903 that filters a video signal input from an external image pickup device 901 such as a television camera (at intervals of a predetermined number of frames), performs A / D conversion.
Stored as image data. The object extraction unit 904 calculates the difference between the background image stored in the background image storage unit 905 and the background image in which the object does not exist and the input current image data, and extracts the changed area, that is, the object area.

The characteristic amount calculation unit 906 calculates the object region and calculates the characteristic amount, for example, the area ratio, the movement amount, the movement speed, the aspect ratio,
Whether or not the moving object is a person is calculated by calculating the center of gravity, the height, etc., and comparing the calculated feature amounts with the object determination unit 907 when the apparatus considers the person as a person. Is determined. As a result, if it is determined that the person is a person, a warning is issued. The display unit 908 also displays a captured image, and displays a warning in the setting menu and at the time of reporting.

Next, a conventional person tracking process in the feature amount calculation unit 906 and the object determination unit 907 will be described according to the procedure of the flowchart shown in FIG. First, in step S1001, the object is extracted by the object extracting unit 904, and then in step S1002, the feature amount calculating unit 906 calculates the feature amount of the object region, for example, the area ratio, the aspect ratio, the moving amount, and the moving speed. Next, in step S1003, the object determination unit 907 compares the calculated feature amount with a specified value when the device considers it to be a person. Then, the tracking process is stopped unless all the feature quantities satisfy the specified values.

In step S1004, it is determined whether the center of gravity of the object area is within the detection area. As a result, if the center of gravity is outside the detection area, the tracking process is stopped. Next, in step S1005, it is determined whether or not the search range of the previously processed frame and the current frame overlap, with the search range being a region obtained by adding several to several tens of pixels in each of the vertical and horizontal directions to the object region of the frame. To do. If they do not overlap, the tracking process is stopped. Step S10
At 06, it is determined whether or not the input current tracking object is the same as the object in the previously processed frame, and if it is determined that they are not the same, the tracking processing is stopped.

If the result of the above determination is that it is the same as the object, the number of times of tracking processing is compared with a specified value in step S1007, and if it has reached the specified value, an alarm is issued in step S1008. On the other hand, if the number of tracking processes has not reached the specified value, the number of tracking processes is added in step S1009, and the process returns to step S1001.

[0008]

However, in the above prior art, even if the extracted object is determined to be a person, the tracking process is immediately stopped when the center of gravity is outside the detection area. Therefore, when the object moves and enters the detection area, the tracking process is started from that point and the number of tracking processes is counted. Therefore,
There was a problem that the reporting was delayed. On the other hand, if the center of gravity enters the detection area without being tracked and the alarm is issued immediately, there is a problem in that the probability of alarming even a moving object other than a person is increased.

Further, when a plurality of persons are extracted close to each other, the extraction frames circumscribing the object are overlapped and combined to recognize the object as a large extraction frame, and as a result, it cannot be determined as a person or is detected. When a person inside the area and a person outside the detection area are photographed close to each other on the screen, they are combined into a large object, and as a result, the center of gravity is outside the detection area, and there is a problem that no alert is issued. .

Furthermore, since the tracking processing interval of the input image data is constant, it is difficult to detect the movement of a person who is far from the image pickup apparatus or a person whose movement is slow, and it is difficult to extract and track the person. There was a point.

The present invention has been made in view of the above circumstances, and when an object that has entered the detection region is determined to be a person, a motion that can promptly and reliably issue an alarm is issued. A tracking device and a person tracking method thereof are provided.

[0012]

According to the present invention, a motion for issuing an alarm when a person is determined as a result of processing an imaged image to extract an object and tracking the object in a plurality of images A tracking device, a person determining means for determining whether the object is a person from the extracted feature amount of the object, and a center of gravity for determining whether the center of gravity of the object is within a predetermined detection area. The extracted object includes position determining means, object coincidence determining means for determining whether the object is the same as the previously tracked object, and tracking number determining means for determining the number of times of tracking processing of the object. When the object is a person and is the same as the previously tracked object even if the center of gravity of the object is outside the predetermined detection area, the object determination means for continuing the tracking process is provided. .

In the above configuration, even if a person is extracted outside the detection area, the tracking processing is continued for the same person, so that the person entering the detection area can be quickly and surely notified. It will be possible.

Further, in the object determining means, the center of gravity of the extracted object is within the detection area, the object is a person, and the object tracked by the previous tracking processing is outside the detection area. In this case, when the object and the object extracted this time are the same, the alarm is issued when the number of tracking processes including the tracking process outside the detection area reaches a predetermined number or more. .

In the above configuration, even when a person is extracted outside the detection area, the tracking process is continued for the same person,
By adding the number of tracking processes, for a person who enters the detection area, immediately or after a small number of tracking processes,
It is possible to issue a report without error.

Area calculating means for calculating the area of an object extraction frame circumscribing the extracted object, counting means for counting the number of persons in the object extraction frame from the shape features of the object, and the object extraction A feature amount calculating means including a ratio calculating means for calculating at least one of an area ratio and an aspect ratio of each object in the object extraction frame based on the area of the frame and the number of persons in the object extraction frame, Characterize.

According to the above configuration, even when a plurality of persons who are close to each other are extracted, it is possible to accurately calculate the feature amount such as the area ratio and the aspect ratio of each person, and it is possible to reliably calculate the person in the detection area. It is possible to issue a report to.

Further, it is characterized by further comprising processing interval setting means for setting a processing interval according to a distance from the image pickup means of the extracted object or a moving speed or a moving distance.

According to the above construction, according to the distance or moving speed or moving distance of the extracted object from the image pickup means,
The tracking process can be performed at an appropriate processing interval, and the accuracy of the tracking process can be improved.

Further, the apparatus is provided with a plurality of background image storage means for respectively storing background images when the object does not exist, and the object determining means is a distance or a moving speed or a moving distance of the extracted object from the image pickup means. Accordingly, the tracking processing is performed at different processing intervals by using the background images stored in the separate background image storage means.

According to the above construction, according to the distance or moving speed or moving distance of the extracted object from the image pickup means,
It is possible to perform tracking processing in parallel at appropriate processing intervals for objects having different distances, etc., and it is possible to improve the accuracy of tracking processing.

According to the present invention, a person tracking method of a motion tracking device for issuing an alarm when a person is judged as a result of processing a captured image to extract an object and tracking the object in a plurality of images. In the case where it is determined that the object is a person from the feature amount of the extracted object, and it is determined that this object is the same as the previously tracked object,
It is characterized by including an object determining step of continuing the tracking process even when the center of gravity of the object is outside the detection region by determining whether the center of gravity is within the predetermined detection region.

In the above procedure, even if a person is extracted outside the detection area, the tracking processing is continued for the same person, so that the person entering the detection area can be quickly and surely notified. Will be possible.

In the object determining step, the center of gravity of the extracted object is within the detection area, the object is a person, and the object tracked by the previous tracking process is outside the detection area. In this case, when the object and the object extracted this time are the same, the alarm is issued when the number of tracking processes including the tracking process outside the detection area reaches a predetermined number or more. .

In the above procedure, even if a person is extracted outside the detection area, if the person is the same, the tracking processing is continued, and the number of times of the tracking processing is added, so that the person who enters the detection area is immediately notified. Alternatively, after a small number of times of tracking processing, it is possible to issue an alarm without error.

The area of the object extraction frame circumscribing the extracted object is calculated, the number of persons in the object extraction frame is counted from the shape feature of the object, and the area of the object extraction frame and the object extraction are calculated. It is characterized by further comprising a feature amount calculating step of calculating at least one of an area ratio and an aspect ratio of each object in the object extraction frame based on the number of persons in the frame.

According to the above procedure, even when a plurality of persons who are close to each other are extracted, it is possible to accurately calculate the feature amount such as the area ratio and the aspect ratio for each person, and it is possible to reliably calculate the person in the detection area. It is possible to issue a report to.

Further, the method is characterized by further comprising a processing interval setting step of setting a processing interval according to a distance or a moving speed or a moving distance of the extracted object from the image pickup means.

According to the above procedure, according to the distance or the moving speed or the moving distance of the extracted object from the image pickup means,
The tracking process can be performed at an appropriate processing interval, and the accuracy of the tracking process can be improved.

Further, there is a background image storing step for separately storing a background image when the object does not exist, and in the object determining step, a distance or a moving speed or a moving distance of the extracted object from the image pickup means is set. Accordingly, the tracking processing is performed at different processing intervals using the background images stored separately.

According to the above procedure, according to the distance or moving speed or moving distance of the extracted object from the image pickup means,
It is possible to perform tracking processing in parallel at appropriate processing intervals for objects having different distances, etc., and it is possible to improve the accuracy of tracking processing.

[0032]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings. [First Embodiment] FIG. 1 is a block diagram showing the functional arrangement of a motion tracking device according to the first embodiment of the present invention. The motion tracking device according to the present embodiment includes an imaging device 101 and an input unit 10.
2, an input image storage unit 103, a shape feature extraction unit 104, an object extraction unit 105, a background image storage unit 106, a feature amount calculation unit 107, an object determination unit 108, a display unit 109.

A video signal photographed by an image pickup device 101 such as a television camera is A / D-converted after a high frequency component is removed by a filtering process in an input unit 102, and an input image is obtained at intervals of a predetermined number of frames. Storage unit 10
3 is stored as image data.

The shape feature extraction unit 104 filters the image data read from the input image storage unit 103 to extract edge information. The object extraction unit 105 stores background image data stored in the background image storage unit 106 in advance when there is no moving object, image data several frames before, and current image data stored in the input image storage unit 103. Are compared to calculate difference data, and binarization processing is performed based on a specific threshold value of the apparatus to extract a brightness change region in the input image data. Further, the binarized data is subjected to expansion / contraction processing for noise removal, and then labeling processing is performed to give the same label to adjacent pixels, thereby separating each connected component into an object. Extract a region.

The feature quantity calculation unit 107 includes a shape feature extraction unit 1
From the edge information of the object region obtained by 04, the circular or elliptical shape feature corresponding to the head of the person and the width feature from the head to the shoulder are calculated, and the area ratio and aspect ratio of the object region to the person are calculated. , The center of gravity, the moving amount, the moving speed, and the like are calculated in the following manner at intervals of a predetermined number of frames.

(1) The area ratio is the number of pixels contained in the object extraction frame obtained by multiplying the number of vertical and horizontal pixels A and B of a frame circumscribing the extracted object region (hereinafter referred to as the object extraction frame). Taking the ratio of S0 and the reference pixel number S1 as a person,
The area ratio is calculated as S0 / S1. (2) The aspect ratio is the number of pixels A in the vertical and horizontal directions of the object extraction frame,
From B, the aspect ratio is calculated as A / B. (3) The center of gravity is obtained by calculating an average value from the orthogonal coordinates of the extracted object area. (4) For the movement amount, the distance that the center of gravity has moved from the position at the start of the tracking process to the farthest position is calculated. (5) For the moving speed, the coordinate of the center of gravity at time t is (X0
, Y0) and the coordinates of the center of gravity moved in the processing interval Δt are (X1, Y1): moving speed = | (X1, Y1)-(X0, Y0) | / Δt
Calculate by

The object determination unit 108 is a feature quantity calculation unit 107.
Each of the feature amounts calculated by the above is compared with a value that the device has as a predetermined specified value corresponding to the case where the moving object is regarded as a person, and it is determined whether the moving object is a person.
If it is determined that the person is a person, a warning is issued. As the notification, a notification is given by voice or display in the device, or an alarm notification is given to a host system or another device. In addition, the display unit 109 displays an imaged detection area, displays a frame in an area determined to be a person, displays a warning when a report is issued, displays a setting menu, and the like.

In the present embodiment, even when the center of gravity of the detected object is outside the detection area, the object determination unit 108 satisfies the specified value corresponding to a person, and the feature amount satisfies the search range in the previous frame and the current value. If the search range in the frame overlaps, tracking processing is performed, and 1 is added to the current number of times of tracking. On the other hand, if the search ranges do not overlap, tracking is stopped. Then, when the number of times of tracking is equal to or more than a specified value that the device has, the moving object is regarded as a person,
When it is determined that the center of gravity is within the detection area and is the same object as the object being tracked outside the detection area, an alarm is issued immediately.

In this case, in the shape feature extraction unit 104, the height of the moving object being tracked outside the detection area, the shape from the head to the shoulder, the aspect ratio, the moving direction, and other feature quantities and color information are used to determine the outside of the detection area. Determines whether the person being tracked and the person in the detection area are the same person.

Next, the procedure of the person tracking process by the motion tracking device of this embodiment will be described in detail with reference to the flowcharts shown in FIGS.

FIG. 2 shows a first example of the person tracking processing procedure. First, in the object extraction process of step S201 and the feature amount calculation process of step S202, the feature regions such as the area ratio, the aspect ratio, the center of gravity, the moving amount, and the moving speed are calculated as described above. Next, in the characteristic amount determination step of step S203, the calculated characteristic amount is compared with a specified value that corresponds to the case where the device is considered to be a person in advance. Then, when all the prescribed values are satisfied, it is determined that the moving object is a person. If there is even one feature value that does not satisfy the specified value, it is determined that the moving object is not a person, and the tracking process is stopped.

Next, in the center-of-gravity position determination step of step S204, it is determined whether or not the center of gravity calculated in the characteristic amount calculation step of step S202 is within the detection area. If the center of gravity is within the detection region, the process proceeds to the first search range determination step of step S205, and the region in which several to several tens of pixels are added to the object region of the frame in each of the vertical and horizontal directions is set as the search range, and It is determined whether or not the search range in the frame is overlapping with the search range of the moving object extracted in the previously processed frame.

If the search ranges overlap, it is considered that tracking has been established, and in the tracking object confirmation step of step S206, the moving object currently being tracked is the same as the moving object tracked in the previously processed frame. Determine if there is. If the result of determination is that the moving objects are the same, in the first tracking number determination step of step S207, the tracking processing number is compared with a predetermined prescribed value, and if it has reached the prescribed value, the reporting step of step S208 Is issued in. If the number of tracking processes has not reached the specified value, 1 is added to the count value of the number of times of tracking in the tracking number counting process of step S209, and the process returns to step S201 to proceed to the object extracting process in the next tracking processing frame.

On the other hand, in the center-of-gravity position determination step of step S204, when the center of gravity is not within the detection area, the process proceeds to the second search range determination step of step S210, and the search range of the moving object in the current frame is the previous frame. It is determined whether or not it overlaps with the search range extracted in. If the result of determination is that the search ranges overlap, it is considered that tracking has been established, and in the second tracking number determination step of step S211, the number of tracking processes is compared with a predetermined specified value.

When the number of times of tracking processing has reached a predetermined specified value, it is judged that the tracking object is a person, but since it is outside the detection area, no alarm is issued and the tracking processing is continued. In order to do so, the process returns to the object extracting step of step S201. If the number of tracking processes has not reached the specified value, 1 is added to the count value of the number of tracking in the tracking number counting process of step S209, and the process returns to the object extracting process of step S201.

FIG. 3 shows a second example of the person tracking processing procedure. The same steps as those in FIG. 2 are designated by the same reference numerals. In this second example, when it is determined that the object is a person in the feature amount determination step of step S203, the search range in the current frame is extracted in the previous frame in the search range determination step of step S205. It is determined whether or not it overlaps with the search range of the object. If the search ranges overlap, the number of tracking processes is compared with a predetermined value in the tracking number determination step of step S207.

If the number of tracking processes has reached the specified value, it is determined in the barycentric position determining step of step S204 whether or not the barycenter is within the detection area. If the result of determination is that the center of gravity is outside the detection area, no alarm is issued. On the other hand, if the center of gravity is within the detection area, step S
In the tracking object confirmation step of 206, it is determined whether the moving object currently being tracked is the same as the moving object being tracked in the previous frame, and if it is the same object, step S
A warning is issued in the warning step 208.

As described above, in the present embodiment, if the feature amount of the moving object is within the range of the specified value as a person even outside the detection area, the tracking process is performed and the tracking count is set to 1
When the center of gravity of the moving object enters the detection area, if the number of times of tracking is greater than the specified value and the same moving object (person) is detected outside the detection area, immediately I am trying to report it. As a result, it is possible to shorten the time from when the person enters the detection area until the notification is issued. Furthermore, it is possible to issue a warning even if the user has entered the detection area for a short time. If the moving object that was tracked outside the detection area is different from the moving object that was detected after entering the detection area, the moving object after entering the detection area is tracked, and as a result, the moving object is When a person is a person and the number of times of pursuit is equal to or more than a specified value, the notification is issued, so that undetected or erroneous notification can be prevented.

Therefore, it becomes possible to continue the tracking process even when a person is extracted outside the detection area.
In addition, since the number of tracking processes can be continuously added, it is possible to reliably and accurately issue an alarm to the same person who has entered the detection area immediately or after a small number of tracking processes.

[Second Embodiment] FIG. 4 is a block diagram showing the functional arrangement of a motion tracking device according to the second embodiment of the present invention. The second embodiment is different from the first embodiment shown in FIG. 1 in the function of the feature amount calculation unit forming a part of the apparatus,
The feature amount can be calculated when a plurality of objects are extracted close to each other. The same components as those in FIG. 1 are designated by the same reference numerals and the description thereof will be omitted.

The feature quantity calculation unit 407 includes a shape feature extraction unit 1
From the edge information of the object area obtained by 04, the shape feature of the person's head which is circular or elliptical and the shoulders following it, the shape feature such as the width of the head and the shoulder width wider than that, the height value, etc. By this, the number of persons in the object extraction frame is estimated. In this case, the shape feature is obtained from the number of pixels of each element. Then, when it is estimated that there are a plurality of persons, the area ratio of the moving object is calculated by dividing the area of the object extraction frame by the number of persons. As a result, it becomes possible to accurately calculate the area ratio.

Next, the procedure of the person tracking process by the motion tracking device of this embodiment will be described in detail with reference to the flowchart shown in FIG. FIG. 5 shows a procedure for calculating the area ratio, which is one of the above-described feature quantities.

First, in the difference data extracting step of step S501, the current image data stored in the input image storage unit 103 and the background image data stored when the moving object stored in the background image storage unit 106 do not exist, , And the difference data is calculated by comparing the image data of several frames before. Next, in the binarization process of step S502, binarization processing is performed on the calculated difference data with reference to a specific threshold value that the device has, and a brightness change region in the input image data is extracted. In the labeling step of step S503, the connected component of the obtained binary data is divided for each area, and a unique label is given to an area in which adjacent pixels are at level 1. Further, in the labeling calculation step of step S504, the number of pixels and the position of the center of gravity of each label determined by the labeling process are calculated.

On the other hand, in the edge data processing step of step S505, the edges of the object are extracted by filtering the input image data. In the shape feature extraction step of step S506, the features of the circular or elliptical edge and the edge extending downward are extracted from the edge data. Further, in the person counting step of step S507, the persons included in the object extraction frame are counted from the edge feature data.

Then, in the area ratio / aspect ratio calculating step of step S508, the area ratio is calculated by dividing the number of pixels of the label obtained as described above by the number of persons. Here, the number to be divided is determined as follows.

When a plurality of persons approach each other, depending on the positions of the centers of gravity of the plurality of persons, (1) the vertical displacement of the center of gravity is 1/4 or less of the number of vertical pixels, and the horizontal displacement is horizontal. When the number of pixels is equal to or more than, the area is ½ of the area of the combined object extraction frame. (2) Vertical shift of center of gravity is 1/4 to 1 / of the number of vertical pixels
If it is 2 and the horizontal shift is equal to or more than the number of horizontal pixels, the area is set to 1 / 1.5 of the area of the combined object extraction frames. (3) When the vertical displacement of the center of gravity is ½ or more of the vertical pixels and the horizontal displacement is the horizontal pixels or more, the area is ¼ of the area of the combined object extraction frames. To do.

Even when there is only one person, the obtained difference data and optical flow
The aspect ratio and the area ratio can be calculated more accurately by setting the value obtained by dividing the number of pixels (label number) of the moving object calculated from the number of pixels in the vertical direction of the extraction area as the width of the object extraction frame. .

As described above, in the second embodiment, even when a plurality of persons approaching each other are extracted, the object extraction frame can be separated as the area of each moving object. Can get closer,
It is possible to prevent a plurality of persons from being detected as a large combined object. Therefore, it is possible to avoid erroneously determining that the person is other than the person, and it is possible to reliably issue the alarm to the person in the detection area.

[Third Embodiment] FIG. 6 is a block diagram showing the functional arrangement of a motion tracking device according to the third embodiment of the present invention. The third embodiment includes a processing interval setting unit 610 in addition to the configuration of the second embodiment in FIG. The same components as those in FIG. 4 are designated by the same reference numerals and the description thereof will be omitted.

The processing interval setting unit 610 has a function of setting the interval of person tracking processing according to the distance or moving speed of a moving object from the image pickup device, and the input unit 102 causes the display unit 109 to display the interval according to the set value. A series of trace processing operations are controlled.

In a motion tracking device for detecting and tracking a person at a distance of 20 m from near and 200 m at far, when a moving object moving in the optical axis direction of the image pickup device is detected from a distance, In this case, the amount of movement of a moving object is small, and it may be difficult to extract difference data and estimate optical flow based on background difference and inter-frame difference. To solve this problem, it is possible to adopt a method of lengthening the interval for processing the input data, but in this case, the difference data is large for a person near the imaging device or a fast motion. This may cause problems such as excessive calculation, a large search range, and an increase in calculation amount.

Therefore, in the present embodiment, when the background image storage unit 106 is one system (for one frame), the moving object is extracted only far from the image pickup apparatus 101, or the moving object having a slow speed is extracted. In this case, the processing interval is set long, for example, every 15 to 50 frames. On the other hand, when the position where the moving object is extracted is close to the imaging device 101, or when the moving object having a high moving speed is extracted, the processing interval is set to be short, for example, 4 to 12 frames. This allows
It is possible to improve the accuracy of detection / tracking of a moving object at a distance and the accuracy of detection / tracking of a moving object that moves fast.
When the moving object is extracted both far and near from the image pickup apparatus 101, the optimum processing interval is set for the processing on the moving object near. As a result, it is possible to improve the accuracy of tracking processing for persons who are nearby.

As described above, in the third embodiment, when the moving object is distant from or near the image pickup apparatus, the person tracking processing interval is set according to the distance from the image pickup apparatus and the moving speed. By setting different values, it is possible to improve the accuracy of person extraction processing and tracking processing.

[Fourth Embodiment] FIG. 7 is a block diagram showing the functional arrangement of a motion tracking device according to the fourth embodiment of the present invention. The fourth embodiment is provided with a second background image storage unit 706 in addition to the configuration of the third embodiment in FIG. The same components as those in FIG. 6 are designated by the same reference numerals and the description thereof will be omitted.

In the fourth embodiment, the background image storage units 106 and 706 for two frames are provided for each one frame.
The extracted moving object is used properly according to the distance from the imaging device 101 and the moving speed. For example, when the distance to the moving object is 150 m or more, or when the moving speed is slow such as 2 km / h or less, the background image storage unit 106 is used, and the tracking processing interval is long, for example, 15 to 5.
Set to 0 frame interval and process. On the other hand, the imaging device 1
When a moving object near which the distance from 01 is 150 m or less is extracted, the second background image storage unit 706 is used,
Processing is performed by setting a short tracking processing interval of 4 to 12 frame intervals. This enables highly accurate difference data extraction,
The optical flow can be estimated, and the accuracy of person extraction processing and tracking processing can be improved.

Further, when a moving object is extracted both far and near from the image pickup apparatus 101, or when both a slow object and a fast object are detected, the two background image storage units 106 and 706 are respectively set. It is possible to perform processing in parallel at different processing intervals by properly using the above. Therefore, it is possible to perform tracking processing in parallel and issue an alarm even for persons having different distances and moving speeds in the detection area.

FIG. 8 is an explanatory view showing a distance setting method in the case of calculating the distance from the image pickup device to the moving object on the photographing screen. On the screen of the detection area photographed by the imaging device 101, the distance from the bottom of the screen to the imaging device is obtained, and the bottom of the screen is used as a reference, for example, 50
The distance on the screen is preset at intervals of about m. And
The distance to the moving object is estimated by comparing the position of the center of gravity of the moving object on the screen with the set distance. With such a procedure, the distance from the imaging device to the moving object can be obtained.

As described above, in the fourth embodiment, by selectively using the plurality of background image storage units according to the distance to the moving object, it is possible to accurately measure the person regardless of the position or moving speed of the object. Can be extracted and traced.

[0069]

As described above, according to the present invention, when an object that has entered the detection area is determined to be a person, it is possible to promptly and surely issue an alarm. Is obtained.

[Brief description of drawings]

FIG. 1 is a block diagram showing a functional configuration of a motion tracking device according to a first embodiment of the present invention.

FIG. 2 is a flowchart showing a first example of a person tracking processing procedure in the motion tracking device of the first embodiment.

FIG. 3 is a flowchart showing a second example of a person tracking processing procedure in the motion tracking device of the first embodiment.

FIG. 4 is a block diagram showing a functional configuration of a motion tracking device according to a second embodiment of the present invention.

FIG. 5 is a flowchart showing a procedure for calculating an area ratio in the motion tracking device according to the second embodiment.

FIG. 6 is a block diagram showing a functional configuration of a motion tracking device according to a third embodiment of the present invention.

FIG. 7 is a block diagram showing a functional configuration of a motion tracking device according to a fourth embodiment of the present invention.

FIG. 8 is an explanatory diagram showing a distance setting method for calculating a distance from an imaging device to a moving object on a shooting screen in the motion tracking device of the fourth embodiment.

FIG. 9 is a block diagram showing a configuration example of a person tracking device according to a conventional technique.

FIG. 10 is a flowchart showing a person tracking processing procedure in a conventional person tracking device.

[Explanation of symbols]

101 imaging device 102 Input section 103 input image storage unit 104 Shape feature extraction unit 105 object extraction unit 106 background image storage unit 107,407 Feature amount calculation unit 108 Object determination unit 109 display 610 Processing interval setting unit 706 Second background image storage unit

─────────────────────────────────────────────────── ─── Continuation of front page (51) Int.Cl. ⁷ identification code FI theme code (reference) G08B 25/00 510 G08B 25/00 510M H04N 7/18 H04N 7/18 K F term (reference) 5B057 AA19 BA02 CA12 CA16 DA07 DA15 DB02 DC06 DC14 DC16 DC32 DC36 5C054 CC05 CG06 CH01 EA01 EA05 EF06 FA09 FC13 FF07 HA18 HA19 5C084 AA02 AA07 AA19 BB04 BB33 CC17 DD12 EE02 FF20 FF20 FF20AFF20AFF20AFF20AFF20AFF20AFF20A04 GG03 GG09 5L096 BA02 CA02 FA06 FA18 FA54 FA69 GA34 HA03 JA11

Claims (10)

[Claims] 1. Processing an imaged image to extract an object,
As a result of tracking processing of the object in a plurality of images, a motion tracking device that issues a warning when it is determined to be a person, and determines whether the object is a person from the feature amount of the extracted object. Person determining means, center of gravity position determining means for determining whether the center of gravity of the object is within a predetermined detection area, and object matching determination for determining whether the object is the same as the previously tracked object Means and a tracking number determination means for determining the number of times of tracking processing of the object, and even if the center of gravity of the extracted object is outside a predetermined detection area, the object is a person and is tracked last time. A motion tracking device characterized by comprising an object determination means for continuing the tracking process when the object is the same as the object. 2. The object determining means has the center of gravity of the extracted object within the detection area, the object is a person, and the object tracked by the previous tracking processing is outside the detection area. In this case, when the object and the object extracted this time are the same, the alarm is issued when the number of tracking processes including the tracking process outside the detection area reaches a predetermined number or more. The motion tracking device according to claim 1. 3. An area calculation means for calculating the area of an object extraction frame circumscribing the extracted object, a counting means for counting the number of persons in the object extraction frame from the shape feature of the object, and the object extraction. A feature amount calculating means including a ratio calculating means for calculating at least one of an area ratio and an aspect ratio of each object in the object extraction frame based on the area of the frame and the number of persons in the object extraction frame, The motion tracking device according to claim 1, wherein the motion tracking device is characterized in that: 4. The processing interval setting means for setting a processing interval according to a distance or a moving speed or a moving distance of the extracted object from the image pickup means, according to any one of claims 1 to 3. The described motion tracking device. 5. A plurality of background image storage means for separately storing background images when the object does not exist,
The object determining means performs tracking processing at different processing intervals using background images stored in separate background image storage means according to the distance or moving speed or moving distance of the extracted object from the imaging means. The motion tracking device according to claim 4, wherein: 6. Processing the captured image to extract an object,
A person tracking method of a motion tracking device that issues a warning when it is determined that the object is a person as a result of tracking processing of the object in a plurality of images, and the object is a person from the feature amount of the extracted object. If it is determined that this object is the same as the previously tracked object, then it is determined whether the center of gravity of the object is within a predetermined detection area and it is outside the detection area. Even if there is, a person tracking method having an object determination step of continuing the tracking process. 7. In the object determining step, the center of gravity of the extracted object is inside the detection area, the object is a person, and the object tracked in the previous tracking process is outside the detection area. In this case, when the object and the object extracted this time are the same, the alarm is issued when the number of tracking processes including the tracking process outside the detection area reaches a predetermined number or more. The person tracking method according to claim 6. 8. The area of an object extraction frame circumscribing the extracted object is calculated, the number of persons in the object extraction frame is counted from the shape feature of the object, and the area of the object extraction frame and the object extraction are calculated. 8. A feature quantity calculating step of calculating at least one of an area ratio and an aspect ratio of each object in the object extraction frame based on the number of people in the frame.
The person tracking method described in. 9. The method according to claim 6, further comprising a processing interval setting step of setting a processing interval according to a distance or a moving speed or a moving distance of the extracted object from the image pickup means. Person tracking method. 10. A background image storing step of separately storing a background image when the object does not exist, wherein in the object determining step, a distance or a moving speed or a moving distance of the extracted object from an image pickup means is set. 10. The person tracking method according to claim 9, wherein the tracking processing is performed at different processing intervals using the background images stored separately.