JP2006012013A - Mobile object tracking device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To sufficiently exhibit the function of mobile object tracking without performing parameter adjustment. <P>SOLUTION: This device comprises a mobile object area extraction part 41 extracting a mobile object area A from an image taken by an ITV camera 200; a parameter acquisition part 42 determining and setting, based on the mobile object area A, an approach detection area B and an object tracking template C necessary for tracking a mobile object; an object tracking part 31 tracking the mobile object in the taken image by use of the template; a brightness environment monitoring part 53 monitoring the change in brightness in the vicinity of a shooting position; a parameter adjustment part 43 re-operating the parameter acquisition part 42 in case that the brightness in the vicinity of the shooting position is changed; an angle-of-view monitoring part 52 monitoring the change in angle of view of the ITV camera 200; a mobile object area monitoring part 51 monitoring the change in number of the mobile object areas A; and a mobile object area adjustment part 44 re-operating the mobile object area extraction part 41 in case that the angle of view or the number of the mobile object areas A is changed. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a moving object tracking device used for grasping, for example, a vehicle traffic flow on a road and a flow of people on the road.

This type of moving object tracking device is a device having a function of recognizing a moving object by template matching from a shooting screen of a camera and individually tracking it, and detecting that a moving object has entered, Object detection means that handles a part or all of the part as an initial template, and a part with a high degree of similarity from the next shooting screen by template matching is recognized as a moving object after moving, and this is repeated to move the moving object It has a basic configuration including an object tracking means for tracking (for example, Patent Document 1).
JP 2002-31890 A

  However, in the case of the above-described conventional example, the parameter adjustment required to be performed at the time of installation of the apparatus is currently performed manually by using a sample image as a procedure, and there is a problem that this operation is troublesome. . In particular, if fluctuation due to disturbance occurs or the characteristics of the target moving object are non-uniform, it takes a long time to adjust the parameters. Moreover, since it is performed manually, the variation between workers is large. Moreover, it is impossible to prepare in advance all sample images that affect the tracking, and it is therefore very difficult to cope with sudden situations such as lightning strikes or lighting breaks. . These points are major obstacles in improving the tracking performance of the apparatus.

  It is also essential to strictly manage the direction of the shooting camera during operation. This is because it is necessary to perform parameter adjustment again when the direction of the photographing camera changes. In addition, it is very difficult to apply this device to a revolving camera. As long as the template matching method is used, the processing load is heavy, and the improvement of the processing speed increases the cost of the device itself. There is another drawback of becoming.

  The present invention was created under the above-described background, and the main object of the present invention is a moving object that can fully perform the original function of the apparatus without manually adjusting the parameters of the template. It is to provide a tracking device.

  A moving object tracking device according to the present invention includes an image memory that sequentially records image data input from a camera that captures a moving object, and a moving object that repeats repetitive movement from a captured image based on the image data on the image memory. A moving object region extracting unit that extracts a moving object region that is a trajectory of the object, a parameter acquiring unit that obtains and sets parameters of an entry detection region and an object tracking template necessary for tracking the moving object based on the moving object region, and , Tracking the moving object in the captured image based on the image data in the image memory using the entry detection area and the object tracking template to obtain the position information, and monitoring the change in brightness near the shooting location A brightness environment monitoring unit, a parameter adjustment unit that re-activates the parameter acquisition unit when there is a change in brightness near the shooting location, and a camera image An angle-of-view monitoring unit that monitors a change in the angle of view of the camera based on an angle-of-view signal input from the operating device, and a moving object region monitoring unit that monitors a change in the number of moving object regions extracted by the moving object region extraction unit And a moving object region adjusting unit that re-activates the moving object region extracting unit and the parameter acquiring unit when a change in the angle of view of the camera or a change in the number of moving object regions appears.

  Preferably, a stop determination unit that determines whether or not the moving object in the captured image is stopped based on image data on the image memory, and that at least the moving object is present in the captured image or the moving object It may be configured to add an output unit for outputting that the output is stopped.

  When extracting the moving object area, the moving object area extraction unit sequentially obtains the time difference image, while the common part between the time difference image of a predetermined time and the current time difference image is determined as the moving object area candidate. The number of candidates for the moving object region candidate for a certain time is obtained for each pixel of the moving object region candidate, the pixel whose number of candidates exceeds the threshold is selected, and the region constituted by the pixel is moved. A structure having a function as an object region may be used.

  Further, when the moving object region extraction unit has a function of obtaining the moving direction of the moving object by optical flow, and the approach detection region and the object tracking template are rectangular, the parameter acquisition unit is related to the parameters of the rectangular approach detection region. The moving object area is divided into a plurality of moving object movement directions, and the position of the approach detection area for starting the tracking of the moving object for each of the divided areas is approximately located on the center line of the moving object area in the moving direction. And setting the width of the approach detection area to be slightly larger than the width of the moving object area, and setting the length of the approach detection area according to the aspect ratio of the moving object area. With respect to the parameters of the object tracking template, the center position of the object tracking template is set to a position approximately coming to the center of the moving object. Same width as the width of the moving object region, the length of the object tracking templates may be used those which is configured to set smaller than its width.

  More preferably, a thinning template obtained by thinning a template image for each predetermined pixel is used as an object tracking template, and a structure having a function of setting the thinning ratio corresponding to the width of the moving object region is used. Is desirable.

  As for the object tracking unit, tracking of the moving object is finished at a position where the width of the moving object area is equal to or smaller than a predetermined value and / or the tracking representative point of the moving object on the moving object area is out of the moving object area. It is sometimes desirable to use a configuration having a function of interrupting tracking of the moving object.

  It also has a function to obtain the movement vector of the moving object being tracked at any time, estimate the next movement destination from the transition of the movement vector, and set the range to search for the moving object to the range centered on the movement destination. It is desirable to use the one having the structure described above.

  As described above, in the case of the moving object tracking device according to the present invention, the parameters of the approach detection area and the object tracking template are automatically adjusted, so that the original function of the device can be sufficiently exerted without manually adjusting parameters. Is possible. In addition, since the parameters are readjusted each time according to a change in brightness near the shooting location, a change in the angle of view of the camera, a change in the situation of the moving object, and the like, the tracking accuracy of the apparatus increases. In addition, it is possible to use a swivel camera, which widens the application range of the apparatus. Therefore, there is a great significance in improving the performance of the apparatus.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram of the moving object tracking device, FIG. 2 is a flowchart of the moving object tracking device, FIG. 3 is an explanatory diagram showing the moving object region extraction algorithm in the flow of image processing, and FIG. 4 shows the extracted moving object region. FIG. 5 is an explanatory diagram for explaining a method for obtaining a moving direction of a moving object on the moving object region, FIG. 6 is a schematic diagram showing a state where an approach detection region is set in the moving object region, and FIG. FIG. 8 is a diagram showing the shape of the object tracking template, FIG. 8 is a schematic diagram showing how the parameters of the object tracking template are set in the moving object region, and FIG. 9 shows the next movement vector predicted from the transition of the movement vector of the moving object. FIG. 10 is a diagram showing a relationship between a prediction vector and a template search range, FIG. 11 is a diagram showing a thinning template, and FIG. 12 is a schematic diagram showing a search using a thinning template.

  The moving object tracking device 100 described here is, for example, a device used to measure the vehicle traffic flow on the road and the flow of people in the passage, and as shown in FIG. The image data output from the ITV camera 200 installed on the ceiling or the like and the angle-of-view signal output from the camera angle-of-view adjuster 300 that adjusts the angle of view of the ITV camera 200 are input. Based on these input data In addition, the moving object photographed by the ITV camera 200 is individually tracked, and functions such as detecting that the moving object exists in the monitoring range and that its movement stops are included.

  The apparatus 100 includes an image capturing unit 10 that periodically captures an image input from the ITV camera 200 and performs A / D conversion, an image memory 20 that sequentially stores image data converted into digital values, A CPU or the like to which image data on the image memory 20 is sequentially input, and its function blocks 30, 40, and 50 that perform various functions by sequentially executing the software, and an output that outputs the detection results as an apparatus to the outside The basic configuration includes the unit 60.

  The functional blocks 30 to 50 are included in the configuration for convenience of explanation, and specifically have the following functions. That is, the functional block 30 includes an object tracking unit 31 and a stop determination unit 32. The functional block 40 includes a moving object region extraction unit 41, a parameter acquisition unit 42, a parameter adjustment unit 43, and a moving object region adjustment unit 44. The functional block 50 includes a moving object region monitoring unit 51, a field angle monitoring unit 52, and a brightness environment monitoring unit 53. A switch and an operation input unit for performing various settings are not shown. Details of each part will be described below.

  The moving object area extraction unit 41 extracts a moving object area A (see FIG. 4) that is a locus of a moving object that repeats repetitive movement from the captured image based on the image data on the image memory 20, and moves the moving object. It is a basic configuration for determining the direction by optical flow. In extracting the moving object area, the time difference image is sequentially obtained, while the common part between the time difference image of a predetermined time and the current time difference image is obtained as a moving object area candidate. For each candidate pixel, the number of times that the candidate has become a candidate for a certain period of time is obtained, the pixels whose number of candidates has exceeded the threshold value are selected, and the region constituted by these pixels is set as the moving object region. ing.

  That is, the extraction of the moving object region A in the captured image is performed by a method of accumulating (voting) a certain amount of pixels on the image memory 20 that are likely to be moving. In detecting a moving object, the detection is performed based on a time difference that is stable against changes in ambient lightness environment. However, if a moving object is detected using only the time difference, it is likely to be affected by shaking, shadows, and image signal noise of non-target objects, so only overlapping portions of successive time differences are voted. In such a processing method, when the moving object projected in the image is moving in the vertical direction, the voting efficiency is good at the back of the screen where the time difference is likely to overlap (upper part in the captured image), On the other hand, it is bad at the front of the screen (lower part of the photographed image), so in order to eliminate this disparity, weighting is performed so that the ratio of one vote is small at the back of the screen and large at the front. Like to do.

  The specific processing flow is as shown in FIG. Here, the image data on the image memory 20 is sequentially read out every 66 ms. These images are represented as input images 1, 2, 3, 4, 5, 6, 7,. Then, image processing is performed on the read image data as follows.

  First, the grayscale image 1 is obtained by subtracting the input image 1 and the input image 2, and then binarization processing is performed to obtain the binary image 1, and noise is removed to obtain the binary image 2. Similar processing is performed on the input image 2 and the input image 3 to obtain a binary image 2 '. Then, the binary image 2 and the binary image 2 ′ are logically obtained to obtain the binary image 3, and then the noise is removed to obtain the binary image 4, and the expansion / contraction process is performed to obtain the binary image 5 Ask for.

  Then, the grayscale image 2 (weighted image) and the binary image 5 prepared in advance for performing the above weighting are ANDed to obtain the grayscale image 33, and then the maximum filter processing is performed to perform the grayscale image. 4 is obtained, and the gray level image 5 is obtained by performing the processing of the minimum filter.

  The brightness value of the gray image 5 generated by the above-described series of processing performed on the input images 1, 2, and 3 is added to the brightness value of the gray image 6 reset at the start of moving object region extraction. The gray image 6 ′ is obtained, and the gray image 5 ′ and the gray image 6 ′ are similarly added to the input images 2, 3 and 4 to obtain the gray image 6 ″. Thereafter, the above processing is sequentially performed on the input images 3, 4, 5,.

  By such processing, only the moving object region A is extracted from the photographed image, and if only the pixels exceeding the predetermined reference are further extracted from the extracted portion, a moving object region image is obtained. The moving object area A shown in the moving object area image indicates an area where the moving object is highly likely to move.

  FIG. 4 shows a moving object region image when the moving object is moving in the vertical direction (vertical direction) in the captured image. This moving object region image is recorded in a main memory (not shown) along with data on the moving direction of the moving object. When there are a plurality of moving object areas A, labeling is performed to distinguish each area, and a number related to labeling is also included in the recording target.

  The moving direction of the moving object is obtained as follows. That is, the pixel point in the moving object region is set as the target of the optical flow (see FIG. 5A), and the components of the moving vector of each point set as the target of the optical flow are obtained and added. 5 (b)), the moving direction of the moving object is obtained based on the combined movement vector obtained by the addition. Here, the moving direction of the moving object whose magnitude of the combined moving vector exceeds a predetermined minimum reference is obtained in four types of upward, rightward, leftward, and downward (see FIG. 5C).

  As a result, whether the moving object is moving in the vertical direction in the captured image (when the ITV camera 200 is arranged in the traveling direction of the moving object) or whether it is moving in the horizontal direction (ITV camera 200). Is arranged in a direction perpendicular to the traveling direction of the moving object).

  Thus, the moving direction of the moving object is directly related to the direction of the ITV camera 200. Therefore, not only when the apparatus is activated, but also when the angle of view signal of the camera angle of view operating device 300 detects a change in the direction of the ITV camera 200, the moving direction of the moving object is obtained each time. Details of this point will be described later.

  The parameter acquisition unit 42 obtains and sets the parameters of the approach detection region B and the object tracking template C necessary for tracking the moving object based on the moving object region A extracted by the moving object region extraction unit 41. It has become. In the object tracking unit 31, a moving object that has entered the rectangular approach detection region B is tracked using the rectangular object tracking template C, and parameter adjustment of the approach detection region B and the object tracking template C is performed by the parameter acquisition unit. 42 is automatically performed. When there are a plurality of moving object areas A, parameter adjustment is performed in the same manner for each area.

  Specifically, the parameter acquisition unit 42 obtains parameters of the entry detection region B and the object tracking template C by the following method. Here, a case where the moving object moves in the vertical direction in the captured image will be described.

  First, regarding the parameters of the approach detection area B, the moving object area A is divided into a plurality of moving objects in the moving direction (the moving object area A is divided into three in the illustrated example), and the divided areas (in FIG. 4) A rectangular approach detection area B (shown as entry detection areas B1, B2, and B3 in FIG. 6) that starts tracking a moving object for each of the divided moving object areas A1, A2, and A3) is centered on the moving object. It is set to a position approximately on the center line of the area A (center line in the same direction as the moving direction of the moving object). Then, the widths of the approach detection areas B1, B2, and B3 (the lengths of the approach detection areas in the lateral direction orthogonal to the moving direction of the moving object) are divided into the widths of the moving object areas A1, A2, and A3 (the moving object The length of each of the entry detection areas B1, B2, and B3 (the length of the entry detection area in the vertical direction that is the same as the movement direction of the moving object) is slightly larger than the length of the divided moving object area in the horizontal direction orthogonal to the movement direction. Is set according to the aspect ratio of the divided moving object areas A1, A2, A3. Here, the lengths of the divided moving object areas A1, A2, and A3 are set to about 3/4 of the respective widths.

  On the other hand, with respect to the parameters of the object tracking template C, the rectangular object tracking template C1 shown in FIGS. 7 and 8 is set to a position where the center position is approximately at the center of the moving object. The width of the object tracking template C1 (the length of the object tracking template in the lateral direction orthogonal to the moving direction of the moving object) is the same as the dimension in the width direction of the moving object region A and along the moving direction of the moving object. For example, the length of the two sides is set to be shorter than the length of the other two sides, for example, to be about ½ of the width.

  The object tracking template C1 shown in FIG. 7A is used when the moving object moves in the vertical direction in the screen, while the object tracking template C2 shown in FIG. Is used when moving in the horizontal direction in the screen, and the aspect ratio is set as shown in FIG.

  When a moving object sequentially enters the entry detection areas B1, B2, and B3, tracking processing is sequentially triggered, but the object tracking template C1 is used for actual tracking. For convenience, the object tracking templates used after entering the entry detection areas B1, B2, and B3 will be represented as object tracking templates C11, C12, and C13.

  By the way, the width of the moving object region A becomes smaller in the moving direction of the moving object as shown in FIGS. This is the same when the moving object moves in the horizontal direction as well as in the vertical direction in the screen. Since the parameters of the object tracking template C1 are set as described above, as shown in FIG. 8, the width of the object tracking template C1 decreases accordingly. That is, the widths of the object tracking templates C11, C12, and C13 are reduced in this order, and as a result, only the boundary portion between the moving object and the background in which many features appear is included. This is exactly the same when the object tracking template C2 is used.

  The parameter acquisition unit 42 also has a function of using a thinning template (see FIG. 11) obtained by thinning a template image for each predetermined pixel as the object tracking template C1, and setting the thinning ratio corresponding to the width of the moving object region A. ing. That is, the thinning rates of the object tracking templates C11, C12, and C13 are set to values that are proportional to the widths of the divided moving object areas A1, A2, and A3, respectively. Since the widths of the divided moving object areas A1, A2, and A3 are reduced in this order as described above, the object tracking template C11 has the maximum thinning rate. This is because the size of the moving object is relatively large on the front side of the screen and the number of pixels of the template is large, the moving vector is large, the search range is wide, and the processing load becomes extremely large. This makes it possible to speed up the processing required for matching. This is a case where the moving object moves in the vertical direction in the screen. When moving in the horizontal direction, if the distance from the ITV camera 200 to the moving object does not change, the thinning rate of the object tracking template C1 used at this time may be basically constant.

  The object tracking unit 31 has a basic configuration in which a moving object in a captured image is tracked by template matching using an approach detection region B and an object tracking template C based on image data on the image memory 20 and position information is obtained. ing.

  Specifically, the moving objects that have sequentially entered the divided moving object areas A1, A2, and A3 are respectively detected by using known detection methods such as time differences using the entry detection areas B1, B2, and B3.

  Thereafter, the moving object is sequentially tracked by template matching using the object tracking templates C11, C12, and C13. For example, when a thinning template as shown in FIG. 12A is used as the object tracking templates C1, C2, and C3, every time this matching is performed, two pixels as shown in FIG. It is performed by moving sequentially by skipping. The same applies to the case where tracking is performed using the object tracking template C2 used when the moving object moves in the horizontal direction in the screen. On the other hand, the position information of the tracking representative point of the moving object is sequentially obtained, and the movement vector of the moving object is obtained from the position change. The position information and movement vector data thus obtained are sequentially stored in a main memory (not shown).

  When there are a plurality of moving object regions A, the processing itself is exactly the same as described above except that tracking of the moving object is performed separately.

  The object tracking unit 31 prevents subsequent moving objects at positions where the width of the moving object region A extracted by the moving object region extracting unit 41 is equal to or less than a predetermined value in order to prevent the tracking of the moving object by mistake. The tracking function is also completed, and when a predetermined tracking representative point of the moving object on the moving object area A deviates from the moving object area A, the tracking of the moving object thereafter is interrupted. Further, in order to increase the speed of the tracking process of the moving object, as shown in FIG. 9, the next moving destination is estimated by obtaining a prediction vector from the transition of the moving vector of the moving object being tracked, as shown in FIG. In addition, it has a function of setting a range for searching for a moving object to a range centered on the destination. That is, when scanning and tracking a moving object by matching using the object tracking template C1 (or C2), the scanning range is limited to an allowable range centered on the next moving destination.

  The brightness environment monitoring unit 53 has a basic configuration having a function of monitoring a change in brightness in the vicinity of the shooting location. Here, the average and standard deviation of the brightness values of the pixels are obtained from the image data on the image memory 20, and it is detected that the brightness in the vicinity of photographing has changed greatly due to day / night, cloudy weather, and the like. Specifically, a threshold is set for the average value and standard deviation of the brightness around the screen center of the captured image (excluding the upper, lower, left, and right 50 dots each), and the daytime and nighttime are mainly divided. Separated. Since there is a lot of light in the daytime and there is also an effect of shadows, it is detected with a high threshold and the sensitivity is lowered. To detect. The result of distinguishing whether it is daytime or nighttime is used to set a threshold value for determining whether or not the moving object has entered the entry detection area at the start of tracking.

  The parameter adjustment unit 43 has a basic configuration in which the detection result of the brightness environment monitoring unit 53 is input, and the parameter acquisition unit 42 is reactivated when a change in the brightness near the shooting location appears. That is, when the brightness near the shooting location changes, the parameter acquisition unit 42 is restarted by the parameter adjustment unit 43, so that the parameters C of the approach detection region B and the object tracking template C are obtained and reset. These parameters are changed to appropriate values according to the current brightness near the shooting location.

  The moving object area monitoring unit 51 has a basic configuration having a function of monitoring a change in the number of moving object areas A extracted by the moving object area extraction unit 41. That is, the current moving object area A is compared with the pixels voted as the moving object area compensation for a certain fixed time in the past, and voting exceeding the threshold is performed in the part that is not registered as the moving object area A. When it is detected that there is a change in the number of moving object areas A, this monitoring result is output to the moving object area adjusting unit 44.

  The view angle monitoring unit 52 has a basic configuration having a function of monitoring a change in the view angle of the camera based on the view angle signal input from the camera view angle operating device 300. The camera angle-of-view operating device 300 is a device that changes the orientation of the ITV camera 200 manually or automatically, and outputs data indicating the elevation angle and the turning angle of the ITV camera 200 as an angle-of-view signal. The ITV camera 200 outputs a zoom-up amount for shooting as a signal (not shown). The angle-of-view monitoring unit 52 constantly monitors the elevation angle, turning angle, and zoom-in amount of the ITV camera 200 through the angle-of-view signal and the like, and outputs these monitoring results to the moving object region adjustment unit 44.

  The moving object region adjustment unit 44 receives the monitoring results of the moving object region monitoring unit 51 and the view angle monitoring unit 52, and changes in the angle of view of the ITV camera 200, the zoom-up amount, or the number of moving object regions A appear. In some cases, the moving object region extraction unit 41 is reactivated. That is, when the change in the angle of view of the ITV camera 200 changes, the moving object area extracting unit 41 is reactivated by the moving object area adjusting unit 44, so that the extraction of the moving object area A and the moving direction of the moving object are obtained. Will be fixed. In this case, the parameter acquisition unit 42 also operates again, and the entry detection area parameter and the object tracking parameter are obtained and reset.

  The stop determination unit 32 has a basic configuration for determining stop of a moving object in a captured image based on image data on the image memory 20. Here, when the velocity component of the moving vector of the moving object obtained by the object tracking unit 31 indicates 0 and the state continues for a predetermined time or more, it is determined that the moving object is stopped.

  The output unit 60 has a basic configuration for externally outputting that a moving object is present in the captured image or that the moving object is stopped. The output unit 60 also includes a signal indicating that there is a moving object that repeats repetitive movement in the captured image when the moving object region A is extracted from the captured image by the moving object region extraction unit 41, and a stop determination unit. 32, when it is determined that the moving object in the captured image is stopped, the tracking data in the memory and the data such as the movement vector are read in accordance with the signal indicating that and the operation of the setting operation unit (not shown). When output, a signal including this data is also output.

  Next, the operation of the moving object tracking apparatus configured as described above will be described with reference to FIG.

  First, when the power switch is turned on, initialization is performed (S1). Then, the image data input from the ITV camera 200 is sequentially recorded in the image memory 20, while the image data on the image memory 20 is sequentially read out every 66 ms. A time difference process and a binarization process are sequentially performed on the read image data, and in this process, the moving object area A in the captured image is extracted and the moving direction of the moving object is obtained (S2).

  Based on the moving object area A and the moving direction of the moving object thus obtained, parameters of the entry detection area B and the object tracking template C are obtained (S3), and the moving object displayed in the captured image is obtained by template matching. Track (S4). Thereafter, it is determined whether or not the moving object is stopped (S5). When it is determined that the moving object is stopped, the fact is output to the outside as a signal (S6). The same applies when the image quality of the ITV camera 200 deteriorates.

  In addition to the process of tracking a moving object (S4, S5, S6), a monitoring process of monitoring changes in brightness near the shooting, changes in the angle of view of the ITV camera 200, changes in the number of moving object areas A, etc. Performed in parallel (S7, S8, S9).

  As a result, when there is no change in brightness or the like in the vicinity of shooting, processing such as tracking of a moving object and monitoring processing are repeatedly performed (S10). On the other hand, when a change in brightness near the photographing occurs, the parameters of the entry detection region B and the object tracking template C are obtained again and adjusted (S2). When a change in the angle of view of the ITV camera 200 occurs, the moving object region A and the moving direction of the moving object are obtained again and adjusted. The same applies when the number of moving object areas A increases. At this time, the fact is output to the outside as a signal, while the moving object area A and the moving direction of the moving object are obtained for the increased amount (S3). Thereafter, processing such as tracking of a moving object and monitoring processing are performed in the same manner as described above.

  Thus, in the case of using the moving object tracking device, the parameters of the entry detection area B and the object tracking template C are automatically adjusted, so that the adjustment work itself becomes unnecessary unlike the conventional case. Moreover, since the parameters are appropriately adjusted according to the width of the moving object area A, etc. regardless of the change in brightness near the photographing, it is possible to accurately track the moving object. Further, since the processing load is reduced by using a thinning template or the like, the processing speed is also increased. Since the parameters are automatically adjusted even if the angle of view of the ITV camera 200 changes, a swivel type can be used as the ITV camera 200, and the application range of the apparatus is widened. In addition, since the processing load of template matching is reduced, template matching can be applied to tracking a large moving object having a large number of pixels, and the degree of freedom with respect to the camera direction is increased. In these respects, it has become possible to achieve a significant increase in performance as a device as compared with the prior art.

Hereinafter, it is a figure for demonstrating embodiment of this invention, Comprising: It is a block diagram of a moving object tracking apparatus. It is a flowchart of the same apparatus. It is explanatory drawing showing the algorithm of moving object area | region extraction with the flow of the image processing. It is a schematic diagram showing the extracted moving object area | region. It is explanatory drawing for demonstrating the method of calculating | requiring the moving direction of the moving object on a moving object area | region. It is a schematic diagram which shows a mode that the parameter of an approach detection area | region is set to a moving object area | region. It is a figure which shows the shape of an object tracking template. It is a schematic diagram which shows a mode that the parameter of an object tracking template is set to a moving object area | region. It is a figure which shows a mode that the next movement vector is estimated from transition of the movement vector of a moving object. It is a figure which shows the relationship between a prediction vector and a template search range. It is a figure which shows a thinning template. It is a schematic diagram which shows the search by a thinning template.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Moving object tracking apparatus 10 Image capture part 20 Image memory 30 Functional block 31 Object tracking part 32 Stop determination part 40 Functional block 41 Moving object area extraction part 42 Parameter acquisition part 43 Parameter adjustment part 44 Moving object area adjustment part 50 Functional block 51 Moving Object Area Monitoring Unit 52 Angle of View Monitoring Unit 53 Brightness Environment Monitoring Unit 60 Output Unit 200 ITV Camera 300 Camera Angle of View Operating Machine

Claims (7)

An image memory that sequentially records image data input from a camera that captures a moving object, and a moving object region that is a locus of the moving object that repeats repetitive movement is extracted from the captured image based on the image data on the image memory. Based on the moving object region extraction unit, the parameter acquisition unit that obtains and sets the parameters of the approach detection region and the object tracking template necessary for tracking the moving object based on the moving object region, and the image data on the image memory An object tracking unit that tracks a moving object in a captured image using the entry detection region and the object tracking template and obtains position information thereof, a brightness environment monitoring unit that monitors a change in brightness near the shooting location, and the vicinity of the shooting location The parameter adjustment unit that re-activates the parameter acquisition unit when a change in the brightness of the image appears, and the angle of view signal input from the camera angle of view An angle-of-view monitoring unit that monitors a change in the angle of view of the camera, a moving object region monitoring unit that monitors a change in the number of moving object regions extracted by the moving object region extraction unit, and at least a change or movement of the angle of view of the camera A moving object tracking device comprising: a moving object region extracting unit and a moving object region adjusting unit that re-activates a parameter acquiring unit when a change in the number of object regions appears. 2. The moving object tracking device according to claim 1, wherein a stop determining unit that determines whether or not the moving object in the captured image is stopped based on image data on an image memory, and at least the moving object is in the captured image. A moving object tracking apparatus comprising: an output unit that externally outputs that an object is present or that a moving object is stopped. 3. The moving object tracking device according to claim 1, wherein the moving object region extraction unit sequentially obtains the time difference image when extracting the moving object region, while the time difference image of a certain time ago and the current time difference image. And the number of times that the candidate number exceeds the threshold for each pixel of the moving object region candidate. A moving object tracking apparatus having a function of selecting and moving a region formed by the pixels into a moving object region. 4. The moving object tracking device according to claim 1, wherein the moving object region extraction unit has a function of obtaining a moving direction of the moving object by optical flow, and the approach detection region and the object tracking template are rectangular. The acquisition unit divides the moving object region into a plurality of moving object movement directions with respect to the parameters of the approach detection region, and the center position of the approach detection region in which tracking of the moving object starts for each divided region is the moving object region Are set to positions approximately on the center line in the moving direction, the width of the entry detection area is slightly larger than the width of the moving object area, and the length of the entry detection area is set according to the aspect ratio of the moving object area. While setting each of the parameters of the object tracking template, the center position of the object tracking template is set to a position that is approximately at the center of the moving object. Body Tracking template width same as the width of the moving object region of the moving object tracking device, characterized in that has a length of the object tracking template configured to be smaller than its width. 5. The moving object tracking device according to claim 1, wherein the object tracking unit ends tracking of the moving object at a position where the width of the moving object area is equal to or less than a predetermined value and / or the moving object area. A moving object tracking device having a function of interrupting tracking of a moving object when a representative tracking point of the moving object deviates from the moving object region. 6. The moving object tracking device according to claim 1, wherein the object tracking unit estimates a next moving destination from a transition of a moving vector of the moving object being tracked, and determines a range for searching for the moving object. A moving object tracking device having a function of setting a range around the moving destination. 7. The moving object tracking device according to claim 1, wherein the parameter acquisition unit uses a thinning template obtained by thinning a template image for each predetermined pixel as an object tracking template, and uses the thinning ratio as a moving object. A moving object tracking device having a function of setting corresponding to the width of an area.

Images