JPH11252587A - Object tracking device - Google Patents

Abstract

(57) [Summary] [PROBLEMS] To accurately and quickly track an object captured in an image over a wide range. SOLUTION: A distance measuring unit 103 measures a distance to an object imaged by stereo image processing, and an object detecting unit 106 detects the object using distance information. The movement position calculation unit 107 extracts information other than the distance information, for example, information of a grayscale image or a ternary edge image from the position of the object detected by the distance information as a template, performs matching, and tracks the object. Since the object is detected based on the distance information, the template can be accurately extracted. In addition, since the tracking is performed using information other than the distance information, a wide range of tracking can be realized.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to an object tracking apparatus, and more particularly to an object tracking apparatus that tracks an object by performing stereo image processing on a captured image.

[0002]

2. Description of the Related Art As a conventional object tracking method, a method described in "Trial of Vehicle Dynamics Measurement" described in the 17th Image Engineering Conference 16-17 (1986) is generally used. In this conventional technique, object tracking is realized by extracting object candidates for each image and associating the extracted object candidates of a plurality of images.

Further, as a method of associating object candidates,
There is one described in JP-A-6-30417. In this method, an object in an image is extracted using several types of grayscale templates stored in advance, and thereafter, the object is tracked by grayscale template matching, and the template is updated each time tracking is performed.

In this prior art, tracking is performed by template matching, and the template is updated each time tracking is performed. Therefore, it is hardly affected by changes in brightness, and accurate tracking of an object can be realized.

As another object tracking method, Japanese Patent Application No. 8-340254 or Japanese Patent Application No. 8-340 is used as a physical distribution measuring device.
No. 255. For example, by acquiring distance information of an imaging space from a plurality of images by stereo image processing, individually detecting a plurality of objects using the distance information, and tracking each of the objects temporally, for example, traffic It measures physical distribution such as quantity.

In this prior art, since distance information is used, an object can be accurately detected without being affected by a change in luminance level caused by an environmental change such as a shadow generated in an image or a reflection of illumination. Can be detected / tracked. Further, it is possible to separately detect and track a plurality of objects that appear to be overlapped in the captured image, using the distance information.

[0007]

However, in the object tracking method described in Japanese Patent Application Nos. 8-340254 and 8-340255, an object is tracked using distance information measured by stereo image processing. However, distance information measurement by stereo image processing has a problem that it is difficult to track a wide range of objects because the distance accuracy measured decreases as the distance from the imaging device increases.

Accordingly, it is a first object of the present invention to provide an object tracking device capable of tracking an object over a wide range.

In the object tracking methods described in Japanese Patent Application Nos. 8-340254 and 8-340255,
Although an object is detected by creating a projection image of the object from distance information measured by stereo image processing, since a projection image of the object is created for each specific region in space, for example, the vehicle in the image In the case of tracking, there is a problem that when a vehicle runs across two adjacent lanes, the vehicle is detected in each lane.

Accordingly, a second object of the present invention is to enable an object existing over two adjacent specific regions to be accurately detected / tracked.

In the object tracking method described in Japanese Patent Application Laid-Open No. Hei 6-30417, an object is tracked by gray-scale template matching. There is a problem that not only does the computational area need to be widened and the amount of computation becomes enormous, but also that the matching accuracy is poor because the template itself changes greatly in a short time. Accordingly, a third object of the present invention is to provide an object tracking device capable of realizing high-speed and accurate object detection / tracking even in a short distance range.

In the object tracking method described in the prior art, Japanese Patent Application Laid-Open No. Hei 6-30417, an object in an image is detected by using several types of grayscale templates stored in advance.
Although the template is extracted, there is a problem that the amount of calculation at the time of matching becomes large because the template size is set larger than the size of the object. Further, there is a case where the object cannot be cut out with the center of the template, and in this case, there is a problem that the object cannot be tracked.

Accordingly, it is a fourth object of the present invention to provide an object tracking apparatus capable of accurately extracting a template required for tracking without excess or deficiency, reducing the amount of calculation at the time of matching, and realizing accurate object tracking. And

In a general conventional object tracking method, a plurality of objects may overlap on an image and a part of the objects may not be visible. In such a case, it is difficult to separate the objects. There is a problem that the tracking of the object cannot be continued.

Therefore, the present invention provides a fifth object tracking apparatus which can continue to track an object even if an overlap occurs.
Subject.

Further, in the object tracking method described in Japanese Patent Application Laid-Open No. Hei 6-30417, an object in an image is tracked using density template matching. However, the density template requires a large amount of calculation for matching. There is. Therefore, a sixth object of the present invention is to provide an object tracking device capable of reducing the amount of calculation at the time of matching and realizing accurate tracking of an object.

Further, in the object tracking method described in the prior art JP-A-6-30417, since a template is cut out from a grayscale image, there is a problem that the amount of calculation at the time of matching becomes large. Then, there is a method of cutting out the template from the binarized edge image in order to reduce the calculation amount, but in this case, the matching accuracy is reduced.

Therefore, it is a seventh object of the present invention to provide an object tracking device capable of reducing the amount of calculation at the time of matching without lowering the matching accuracy.

In the general prior art object tracking method, the occupancy of the object on the two-dimensional image plane is obtained as the space occupancy of the object in the imaging space. When taking a picture, there is a problem that an error increases. As a countermeasure against this problem, Japanese Patent Application No. Hei 8-3
In Japanese Patent No. 40255, an object detected by stereo image processing is projected on a plane formed by the direction of travel of the object and the direction perpendicular to the observation plane, and the space occupancy is calculated from the projected image. However, this method has a problem that the accuracy can be accurately calculated in a short distance range, but the accuracy is reduced in a long distance range.

Therefore, the present invention provides an object tracking apparatus capable of accurately calculating the space occupancy ratio even when photographing a bird's-eye view or photographing a long distance range.
Subject.

[0021]

In order to solve the above-mentioned first problem, an object tracking apparatus according to the present invention has the following configuration.

That is, a plurality of image pickup units are arranged at predetermined intervals to pick up an image of an image pickup space. The distance measuring unit measures distance information of an imaging space from a plurality of captured images by stereo image processing. The distance information storage unit stores distance information at a reference time. From the distance information obtained from the distance measurement unit and the distance information at the reference time stored in the distance information storage unit, the object projection unit is configured for each specific region in space on a plane formed by the direction of travel of the object and the direction perpendicular to the observation plane. Of the object image on the observation surface. The object detection unit individually detects a plurality of object regions from the object projection image obtained by the object projection unit. The movement position calculation unit calculates a movement position of the object by performing matching at a certain time and a next time using a feature amount other than the distance information for each object detected by the object detection unit. The object information calculating unit calculates the number of objects passing through a predetermined position, the moving speed of the objects, the size of the objects, and the abnormal movement of the objects from the result obtained by the moving position calculating unit.

Further, in order to solve the second problem, in the object tracking device according to the present invention, in the object detection unit according to claim 1, the width calculation unit individually calculates the width of each detected object. The same object determination unit determines whether the objects detected in two adjacent specific regions are the same object based on the width of the object calculated by the width calculation unit.

Further, in order to solve the third problem, the object tracking device of the present invention has the following configuration. That is, in the moving position calculation unit according to the first aspect, the distance determination unit determines whether the position of the object detected by the object detection unit is near or far from the imaging unit. The short-range moving position calculating unit calculates a moving position of the object by performing matching between a time at which the distance information is used as a feature amount and a next time within a range of a short distance from the imaging unit. The long-distance range movement position calculation unit calculates a movement position of the object by performing matching at a certain time and a next time using a feature amount other than the distance information in a range far from the imaging unit.

In order to solve the fourth problem, the object tracking device of the present invention has the following configuration. That is, in the moving position calculating section according to the first aspect or the long range moving position calculating section according to the third aspect, the object characteristic portion extracting section is configured to detect a characteristic portion of the object from the object projection image created by the object projecting section. Is extracted. The characteristic part imaging position calculation unit calculates a position where the object characteristic part extracted by the object characteristic part extraction unit is captured in an image. The template extracting unit extracts information other than the distance information as a template from the position calculated by the characteristic part imaging position calculating unit. The template storage unit stores the template extracted by the template extraction unit. The template matching unit performs matching at a certain time and the next time using the template stored in the template storage unit, and calculates a moving position of the object.

Further, in order to solve the fifth problem, the object tracking device of the present invention has the following configuration. That is, in the template matching unit according to the fourth aspect, the overlap determination unit determines the overlap of the objects on the image from the positions of the plurality of objects detected by the object detection unit.
The template division unit divides the template into a plurality of regions when the overlap determination unit determines that a plurality of objects overlap. The region matching unit performs matching for each region divided by the template dividing unit. The similarity calculating unit calculates the similarity of the entire template by weighting the result obtained by the region matching unit.
The moving position determining unit determines a moving position of the object from the result calculated by the similarity calculating unit.

Further, in order to solve the sixth problem, the object tracking device of the present invention has the following configuration. That is, in the moving position calculating unit according to claim 1 or the moving distance calculating unit according to claim 3, the object imaging position calculating unit calculates a position of the object on the image from the position of the object detected by the object detecting unit. Is calculated. The edge histogram creating unit creates an edge image for the image capturing position calculated by the object image capturing position calculating unit, and creates an edge histogram by scanning in a direction perpendicular to the moving direction of the object. The edge histogram storage unit stores the edge histogram created by the edge histogram creation unit. The edge histogram matching unit performs matching at a certain time and a next time using the edge histogram stored in the edge histogram storage unit, and calculates a moving position of the object.

Further, in order to solve the seventh problem, the object tracking device of the present invention has the following configuration. That is, in the distance measuring unit according to the first aspect, the edge image creating unit creates a ternary edge image from an image captured by the imaging unit. The associating unit measures distance information of the imaging space by associating a plurality of ternary edge images created by the ternary edge image creating unit. In the template extracting unit according to the fourth aspect or the edge histogram creating unit according to the sixth aspect, a template is extracted or an edge histogram is created using the ternary edge image created by the edge image creating unit.

Further, in order to solve the eighth problem, the object tracking device of the present invention has the following configuration. That is, in the object detection unit according to claim 1, the object length calculation unit calculates the length of the object from the object projection image created by the object projection unit, and the object length storage unit calculates the object length by the object length calculation unit for each object. The stored object lengths are individually stored. Further, in the object information calculation unit according to claim 1, the length information of the object stored in the object length storage unit and the movement position calculation unit according to claim 1 or the long distance range movement position calculation unit according to claim 3. The space occupancy of the object is calculated using the position information of the object calculated by the above.

[0030]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The invention according to claim 1 of the present invention provides a plurality of image pickup units arranged at a predetermined interval, and a plurality of images picked up by the image pickup units in an image pickup space by stereo image processing. A distance measurement unit that measures distance information, a distance information storage unit that stores distance information at a reference time, and distance information obtained from the distance measurement unit and distance information at a reference time that is stored in the distance information storage unit. An object projection unit that projects an object image on an observation plane for each specific region in space on a plane defined by the direction of travel of the object and a direction perpendicular to the observation plane, and a plurality of object projection images obtained by the object projection unit. An object detection unit that individually detects the region of the object, and performing a matching at a certain time and the next time using a feature amount other than the distance information for each object detected by the object detection unit to determine a moving position of the object. calculate A moving position calculating unit, and an object information calculating unit that calculates the number of objects passing through a predetermined position from the result selected by the moving position calculating unit, the moving speed of the object, the size of the object, and the abnormal movement of the object. The object is detected using distance information measured by stereo image processing, and the object is tracked using a feature amount other than the distance information, so that the object can be detected / tracked with high accuracy over a wide range. Has an action.

According to a second aspect of the present invention, in the object tracking device according to the first aspect, the moving position calculating section calculates a width of each detected object individually;
It also has the same object determination unit that determines whether the objects detected in two specific regions adjacent to each other from the width of the object calculated by the width calculation unit are the same object. Since it is determined whether the objects detected in the area are the same object, it has an effect that an object existing over two adjacent specific areas can be accurately detected / tracked.

According to a third aspect of the present invention, in the object tracking device according to the first or second aspect, the moving position calculating section determines whether the position of the object detected by the object detecting section is a short distance from the imaging section. A distance determination unit that determines whether the object is a long distance, and a short-range movement position calculation that calculates a movement position of the object by performing matching at a certain time and a next time using the distance information as a feature amount in a short distance range from the imaging unit. And a long-distance range movement position calculation unit that calculates a movement position of the object by performing matching at a certain time and a next time using a feature amount other than the distance information in a long distance range from the imaging unit. Since the object detection / tracking is performed using the distance information measured by the stereo pixel processing in the short distance range, there is an effect that the object can be detected / tracked with high speed and accuracy even in the short distance range.

According to a fourth aspect of the present invention, in the object tracking device according to any one of the first to third aspects, the moving position calculating section or the long distance moving position calculating section is created by the object projecting section. An object feature portion extraction unit for extracting a feature portion of the object from the extracted object projection image, and a feature portion imaging position for calculating a position where the object feature portion extracted by the object feature portion extraction unit is captured in an image A calculation unit, a template extraction unit that extracts information other than distance information as a template from the position calculated by the characteristic part imaging position calculation unit, a template storage unit that stores the template extracted by the template extraction unit, A template that calculates a moving position of an object by performing matching at a certain time and the next time using the template stored in the template storage unit. It consists of a rate matching unit, detects objects based on distance information measured by stereo image processing, and extracts templates.Therefore, templates can be accurately extracted without excess or deficiency, and the amount of calculation at the time of matching can be reduced, with high accuracy. It has the effect that an object can be tracked.

According to a fifth aspect of the present invention, in the object tracking device according to the fourth aspect, the template matching unit overlaps the objects on the image from the positions of the plurality of objects detected by the object detection unit. An overlap determination unit that determines whether a plurality of objects are overlapped by the overlap determination unit, and a template division unit that divides the template into a plurality of regions, and for each of the regions divided by the template division unit. A region matching unit for performing matching, a similarity calculating unit for calculating the similarity of the entire template by weighting the result obtained by the region matching unit, and moving an object based on the result calculated by the similarity calculating unit It consists of a moving position determining unit that determines the position. Predicting a portion to disappear, so to reduce the weight of the similarity of the template of this portion, such an action overlap can track an object be continued occurs.

The invention according to claim 6 is the invention according to claims 1 to 5
In the object tracking device according to any one of the above, the moving position calculating unit or the long distance moving position calculating unit calculates an object imaging position calculating unit that calculates a position on the image of the object from the position of the object detected by the object detecting unit. And an edge histogram creating unit that creates an edge image for the imaging position calculated by the object imaging position calculation unit and creates an edge histogram by scanning in a direction perpendicular to the moving direction of the object; and the edge histogram creating unit An edge histogram storage unit that stores the created edge histogram, and an edge histogram matching unit that performs matching at a certain time and the next time using the edge histogram stored in the edge histogram storage unit to calculate a moving position of the object. And the edge histogram for the object area is correct. Since no can be used to track and create accurately, it can be reduced and accurate object tracking computation amount at the time of matching.

The invention according to claim 7 is the invention according to claims 1 to 6
In the object tracking device according to any one of the above, the distance information measurement unit includes a ternary edge image creation unit that creates a ternary edge image from an image captured by the imaging unit, and a ternary edge image creation unit. And an associating unit that measures distance information of the imaging space by associating the plurality of created ternary edge images with each other.
The valued edge image is used by the template extracting unit or the edge histogram creating unit, and is created when the distance information is measured by stereo image processing.
Since the matching is performed using the template extracted from the binarization, an operation amount of the matching can be reduced.

[0037] The invention described in claim 8 is the first to seventh aspects.
In the object tracking device according to any one of, an object detection unit, an object length calculation unit that calculates the length of the object from the object projection image created by the object projection unit, for each object by the object length calculation unit An object length storage unit for individually storing the calculated object lengths, wherein in the object information calculation unit, the object length information stored in the object length storage unit and a movement position calculation unit or a long distance range movement position The space occupancy of the object is calculated by using the position information of the object calculated by the calculation unit. Since the space occupancy is calculated by using the length information of the object measured by stereo image processing, It has the effect that the space occupancy can be calculated accurately even when taking a picture or when taking a picture far away.

An embodiment of an object tracking device according to the present invention will be described below with reference to the drawings. In the following (first embodiment) to (eighth embodiment), in order to make the description easy to understand, description will be made assuming a traffic flow measuring device as a specific application example.

(First Embodiment) FIG. 1 is a block diagram showing an object tracking apparatus according to a first embodiment of the present invention. In FIG. 1, 101 and 102 are imaging units arranged at a predetermined interval, 103 is a distance measuring unit that measures a distance from an image input by the imaging units 101 and 102 to an object being captured, and 104 is a reference time. A distance information storage unit 105 stores distance information obtained at t0, and stores distance information obtained at a time after the reference time t0 from an imaging plane and a three-dimensional coordinate system representing a distance from the imaging plane to a vehicle from a road surface. And an object projecting unit for projecting an object onto a plane defined by a traveling direction and a perpendicular direction of an observation plane, and an object projecting unit 106 is created by the object projecting means 105. An object detection unit 107 for detecting a region of the object from the projection image performs matching at one time and the next time using a feature amount other than the distance information for each object detected by the object detection unit 106, and A moving position calculating unit 108 for calculating a moving position is based on the result obtained by the moving position calculating unit 107, the number of objects passing through a predetermined position, the moving speed of the object, the size of the object, the space occupancy of the object and the object. Is an object information calculation unit that calculates an abnormal movement of the object.

Next, the operation of the above embodiment will be described. The imaging unit 101 and the imaging unit 10 arranged at a predetermined interval
2 is sent to the distance measurement unit 103, and as shown in FIG. 2, the distance to the object being imaged is divided into a plurality of rectangular regions 203 obtained by dividing the image 201 in a horizontal direction and a vertical direction. Is measured. For a method of measuring distance by stereo image processing, see, for example, Michiyoshi et al. 169-172 (1992-
10) and those described in JP-A-8-294143 are known. In the distance measuring unit 103, the distance information K (X, Y) for each rectangular area is obtained by the method introduced above.
[However, 0 <X ≦ M, 0 <Y ≦ N] is calculated. For the distance information measured from the images captured by the imaging units 101 and 102 at the measurement reference time t0, the position of the plane captured by a method such as the least square method is estimated using information on the measured rectangular area, After complementing the distance of the rectangular area for which the distance could not be measured, the distance information is output to the distance information storage unit 104 as the distance information at the reference time t0.

Measurement time t (t> t) after measurement reference time t0
The distance information measured from the images captured by the imaging units 101 and 102 at t0)
5 is output.

In the object projection unit 105, the distance measurement unit 103
Using the distance information measured by the distance information and the distance information at the reference time t0 stored in the distance information storage unit 104,
The three-dimensional information of the imaging surface and the distance from the imaging surface to the object is converted into three-dimensional information of the road surface and the height of the object from the road surface.

An example of the coordinate conversion operation will be described with reference to FIG. FIG. 3 shows a coordinate system (X, Y, K) indicating the distance from the imaging surface 301 to the vehicle from the imaging surface 301, and a coordinate system (U, V, K) indicating the height of the vehicle from the road surface 302 and the road surface 302.
H).

To calculate the height H of the vehicle imaged in the rectangular area (X, Y), the distance of each rectangular area measured by the distance measuring unit 103 is K (X, Y), and the rectangle at the reference time t0 is calculated. Assuming that the distance for each area is K0 (X, Y), the distance can be obtained by Expression 1. Here, Hc represents the installation height of the camera.

[0045]

(Equation 1) ... (Equation 1)

Further, the position V in the traveling direction of the vehicle imaged in this rectangular area can be obtained by equation (2). Here, θ represents the depression angle of the camera.

[0047]

(Equation 2) … (Equation 2)

The position of the vehicle in the U-axis direction can be obtained by Expression 3 when the U-axis and the X-axis are installed in parallel. However, a and b are determined by the camera to be used and the installation position. U = aX + b (Equation 3)

As described above, by using (Equation 1) to (Equation 3), the coordinate system (X, Y, K) representing the distance from the imaging surface 301 to the vehicle can be converted to the road surface 302.
And a coordinate system (U,
V, H).

Further, in the object projection unit 105, the vehicle imaged in each rectangular area 203 in the image 201 shown in FIG. 2 travels in any specific area (lane in the traffic flow measuring device) in the space. The vehicle information in the real space coordinate system calculated by the coordinate transformation is projected for each lane on a plane (VH plane) formed by the traveling direction of the vehicle and the direction perpendicular to the road surface, and Create a projection image. FIG. 4 shows an example of the projected image of the vehicle created in this way.

In the object detection unit 106, the object projection unit 105
The vehicle region 402 is determined from the projected image 401 of the vehicle for each lane output from the vehicle. As an example of a method of determining the vehicle region 402, there is a method in which a vehicle is determined to be a vehicle if the vehicle projection image has a certain size or more in the V axis direction, and this region is set as a vehicle region. Further, the head position or the rear end position Vr of the vehicle is detected from the vehicle projection image 401. Here, the imaging units 101, 1
When 02 is photographed from the front of the vehicle, the head position is detected, and when photographed from behind, the rear end position is detected. In the following embodiments, the case where the rear end position of the vehicle is detected will be described. However, the case where the head position is detected can be realized by a similar method.

Next, the moving position calculation unit 107 determines an area where the vehicle is imaged on the image captured by the imaging unit 101 from the vehicle area 402 determined by the object detection unit 106, and determines the entire vehicle area. Alternatively, a feature amount other than the distance information is extracted from a part.

An example of a method for calculating a region where the vehicle region 402 is captured on an image will be described with reference to FIG. In FIG. 5, assuming that the rear end of the vehicle can be detected at V = Vr, the road surface 50 at the position of V = Vr from the imaging unit 101 is detected.
The distance K = Kr in the optical axis direction up to 1 can be calculated by Expression 4.

[0054]

(Equation 3) … (Equation 4)

The lane 50 in which the vehicle is traveling
In the range where 2 is captured on the image, an area Y = Y on the image where the distance in the optical axis direction to the road surface is K = Kr
Calculate r. The relationship between the distance K in the optical axis direction and Y on the image is based on the reference time t stored in the distance information storage unit 104.
It can be known by referring to the distance information at 0. FIG.
In the above, if the X axis and the U axis are parallel, the relationship between K and Y can be expressed by Expression 5. Here, A, B, and C are constants determined by camera parameters.

K = A / (B + CY) (Equation 5) Therefore, Yr can be calculated from Kr using Equation 5.

Further, an edge image is created in the range of Y <Yr in the image, and the lane 50 in which the vehicle is traveling is displayed.
In the range where the image No. 2 is imaged, the X axis 503 and the Y axis 505
The edge histograms 504 and 506 are created above, and the edge histograms 504 and 506 exceed the predetermined thresholds THx and THy in a range Xl <X <Xr, Ys <
It is determined that Y <Yr is a part of the area where the vehicle is imaged.

Then, a template is extracted from this area as a feature amount other than the distance information used for tracking. Note that the feature amount extracted as a template includes image luminance information,
Alternatively, differential information, color information, or a combination thereof may be used.

Using the template extracted in this way, matching is performed to calculate the moving position of the object. The template matching calculates a similarity between a template extracted from an image at a certain time t and a nearby area on the image at the next time t + Δt, and a portion corresponding to the template of the object moves to an area having the highest similarity. It is assumed that you have done. For example, when the luminance information of an image is used as a template, the difference sum S
Can be sought. Here, the luminance information of the image at time t + Δt is G, the luminance information of the template is GT,
Let the template size be m × n.

[0060]

(Equation 4) … (Equation 6)

It can be determined that the similarity is the largest in the area where the difference sum S is the smallest. In addition, since the template is two-dimensional image information and changes every moment, after the template matching is completed, it is preferable to replace the template in the most matching area finally determined by the template matching with the past template.

Finally, the object information calculating section 108 calculates the number of objects passing through a predetermined position from the result obtained by the moving position calculating section 107, and calculates the moving speed from the moving amount of the object and the moving time. Then, the size of the object is calculated from the projected image of the object, and abnormal movement of the object is detected from the moving direction and the moving amount of the object.

As described above, according to the first embodiment of the present invention, it is possible to accurately track an object not only in a relatively short range but also in a long range as in the related art.

(Second Embodiment) A second embodiment of the present invention has the configuration shown in FIG. 6, 101 to 108 are the same as those described in the first embodiment and the second embodiment. 601
Is a width calculation unit that individually calculates the width of the object at the moving position of each object detected by the object detection unit 106;
602 determines whether the objects existing in two adjacent lanes are the same object based on the movement position of each object calculated by the object detection unit 106 and the width of each object calculated by the width calculation unit 601. The same object determination unit.

Next, the operation of the above embodiment will be described. 101 to 106 perform operations similar to those described in the first embodiment. The width calculator 601 calculates the width of the object detected by the object detector 106. The operation of the width calculator 601 will be described with reference to FIG.

At a certain time, two lanes 70 adjacent to a vehicle
It is assumed that the vehicle is traveling over the vehicles 1 and 702. At this time, the object detection unit 106 detects the vehicle at the position of Y = Yr on the lanes 701 and 702, respectively. Then, an edge histogram 704 is created on the X axis 703 by the same operation as that described in the movement position calculation unit 107 of the first embodiment, and the vehicle width on the image is calculated. In FIG. 7, the vehicle width on the image of the vehicle detected on the lane 701 is (X2-X1), and the vehicle width detected on the lane 702 is (X3-X2). Here, assuming that the distance in the optical axis direction of the road surface corresponding to Y = Yr is K = Kr, the vehicle width on the image is XW, and the width of the imaging range in the X-axis direction on the image is XN. Of vehicle width W can be calculated by equation (7). However,
C is a constant determined by the camera parameters.

[0067]

(Equation 5) … (Equation 7)

The actual vehicle widths of a plurality of vehicles detected by the object detection unit 106 are individually calculated using equation (7).
Further, in the same object determination unit 602, the width calculation unit 601
From the actual vehicle width calculated in the above, it is determined whether the vehicles detected in two adjacent lanes are the same. When a vehicle is detected at the same Y coordinate position of two adjacent lanes or at a nearby Y coordinate position, the total of the actual vehicle widths of the vehicles detected in the two lanes is calculated. Since the vehicle width of a general vehicle is about 2 m, if the sum of the vehicle widths detected in two lanes is about 2 m, the vehicles can be determined to be the same vehicle. It can be determined that two separate vehicles are running in parallel.

If it is determined that the vehicles detected in the two adjacent lanes are the same, the moving position calculating unit 10
In step 7, the moving position of the vehicle detected in one of the lanes is calculated, and the vehicle detected in the other lane is treated as not being detected.

As described above, according to the second embodiment of the present invention, an object detected in two adjacent lanes is determined to be the same vehicle based on the width of the object, so that the vehicle travels over the two lanes. Vehicles can be accurately detected and tracked.

(Third Embodiment) FIG. 8 is a block diagram of a traffic flow measurement using an object tracking device according to a third embodiment of the present invention. 8, 101 to 101
106 and 108 are the same as those described in the first embodiment. Reference numeral 801 denotes a distance determination unit that determines whether the object detected by the object detection unit 106 is in a range close to or far from the imaging unit, and 802 indicates a movement position of the object when the object is in a range close to the imaging unit. A short distance range movement position calculation unit 803 to be determined is a long distance range movement position calculation unit that determines the movement position of the object when the object is far from the imaging unit.

Next, the operation of the above embodiment will be described. 101 to 106 perform the same operations as those described in the first embodiment. The distance determination unit 801 determines whether the vehicle exists in the short distance range 902 or the long distance range 903 in the imaging range 901 in FIG. The operation of the calculation unit 802 is performed, and if it is in the long distance range 903, the operation of the short distance range movement position calculation unit 803 is performed.

The operation of the short range movement position calculation unit 802 will be described with reference to FIG. The projected image 1001 of the vehicle created at a certain time is represented by H0 (V) and the vehicle rear end position 10
02 is V0, and the projected image 1 of the vehicle created at the next time
003 is H1 (V), and the vehicle rear end position 1004 is V1.

From the projected image H0 (V) at a certain time, V = V0
The nearby projected image Hp (V) is cut out as a partial projected image 1005, and the projected image 1003 of the vehicle at the next time is compared with the projected image near V = V1 to calculate the correlation.

A method using a difference sum will be described as an example of a correlation calculation method. However, the present invention is not limited to the calculation method described below. First, the partial projection image 1
005, and the rear end position 1004 of the projected image 1003 at the next time are summed up.
(V) is calculated. The difference sum S (V) can be calculated by Expression 8. Here, Vs is the size of the partial projection image in the V-axis direction.

[0076]

(Equation 6) … (Equation 8)

Next, the rear end position 1 in the partial projection image 1005
The difference sum S (V) is sequentially calculated while shifting the position of 006 by ΔV in the direction of V <V1. This operation is performed until the amount of deviation of the rear end position exceeds a predetermined amount Ve. Further, the same operation is performed in the direction of V> V1. Then, the position where the obtained difference sum S (V) becomes the smallest is determined as the position having the highest correlation, and this position is determined as the true vehicle movement position.

The vehicle is tracked by the above-described operation, and when the vehicle moves from the short-distance range 902 to the long-distance range 903, the operation of the short-distance range movement position calculation unit 802 ends and the long-distance range movement position is calculated. The operation of the calculation unit 803 starts. The operation of 803 is the same as that described in the first embodiment.

As described above, according to the third embodiment of the present invention, the object is tracked using the distance information measured by the stereo image processing in the short distance range, so that the feature matching at the time of tracking is performed. Can be reduced, and tracking of an object over a wide range can be realized at high speed and with high accuracy.

(Fourth Embodiment) In a fourth embodiment of the present invention, the movement position calculation unit 107 described in the first embodiment or the long distance range movement described in the third embodiment is used. The position calculation unit 803 has the configuration shown in FIG. In FIG. 11, reference numeral 1101 denotes an object characteristic portion extraction unit that determines a characteristic portion of an object based on information of the object region detected by the object detection unit 106, and reference numeral 1102 denotes a characteristic determined by the object characteristic portion extraction unit 1101. A characteristic part imaging position calculation unit 1103 that calculates a position in the image obtained by the imaging unit 101 at which the part is imaged, extracts a template based on the position calculated by the characteristic part imaging position calculation unit 1102. A template extracting unit 1104 stores a template extracted by the template extracting unit 1103.
A template matching unit 105 performs template matching using the template stored in the template storage unit 1104.

Next, the operation of the above embodiment will be described. First, the object characteristic part extraction unit 1101 determines a characteristic part of the object extracted from the vehicle projection image 401 in the vehicle area 402 determined by the object detection unit 106.
When the object is a vehicle, for example, it can be determined that the vehicle rear portion having a number plate, a tail lamp, and the like is a characteristic portion. Therefore, in FIG. Then, Hr and the position Vr in the V-axis direction are calculated.

Next, the characteristic portion imaging position calculation unit 1102 calculates the position where the vehicle characteristic unit 1201 is imaged in the image captured by the imaging unit 101. In FIG. 13, the vehicle feature 1201 has Vr <V of the road surface 1301.
The image is captured at a position on the screen corresponding to the range of <Vt. At this time, Vt can be calculated by Expression 9.

[0083]

(Equation 7) … (Equation 9)

Here, Vr <V <Vt of the road surface 1301
Range Kb <in the optical axis direction from the imaging unit 101 in the range
K <Kt can be calculated by Expression 4 described in the first embodiment. Further, the range Yt <Y <Yb in the Y-axis direction in which the vehicle characteristic portion 1001 is captured on the image can be calculated by Expression 5.

Then, an edge image is created in the range of Yt <Y <Yb in the image, and an edge histogram is created on the X-axis 1303 within the range where the lane 1302 where the vehicle is traveling is imaged. Then, the range Xl <X <Xr in which the vehicle feature section 1201 is imaged is determined. In the template extraction unit 1103, the characteristic part imaging position calculation unit 1
The position (X
A template used for tracking is extracted from 1 <X <Xr, Yt <Y <Yb) and output to the template storage unit 1104.

Finally, the template matching unit 110
In step 5, matching is performed using the template stored in the template storage unit 1104 at a time after the time when the template is extracted, and the moving position of the object is calculated.

As described above, according to the fourth embodiment of the present invention, a template necessary for tracking can be accurately extracted without fail, so that the amount of calculation at the time of matching and the object tracking with high accuracy are realized. it can.

(Fifth Embodiment) In the fifth embodiment of the present invention, the template matching unit 1105 described in the fourth embodiment has the configuration shown in FIG. In FIG. 14, reference numeral 1401 denotes an overlap determination unit that determines overlapping of objects on an image from positions of the plurality of objects detected by the object detection unit 106, and 1402 determines that a plurality of objects overlap with each other by the overlap determination unit 1401. In this case, a template dividing unit 1403 that divides the template into a plurality of regions when performing the matching is a region matching unit that performs matching for each region divided by the template dividing unit 1402, and 1404 weights the result obtained by the region matching unit 1403. The similarity calculating unit 1405 calculates the similarity of the entire template, and 1405 denotes a similarity calculating unit 1404.
Is a moving position determining unit that determines the moving position of the object from the result calculated by

Next, the operation of the above embodiment will be described. First, in a case where a plurality of vehicles are traveling in one lane, a part of a portion corresponding to a template of a vehicle being tracked is hidden by a vehicle traveling behind, and the overlap determination unit 1401 hides a part of the image on the image. Determine if it becomes invisible. FIG.
At a certain time t, a certain lane is moved to two vehicles 150.
1, 1502 are traveling, the rear end position of the vehicle 1501 is V1, the vehicle height is H1, the rear end position of the vehicle 1502 is V2,
Assume that the vehicle height is H2 and the vehicle length is L2. At this time,
When Equation 10 is satisfied, the vehicle feature 1201 of the vehicle 1501
Hides and disappears.

[0090]

(Equation 8) ... (Equation 10)

The height H1B of the hidden area 1503
Can be calculated from Expression 11.

[0092]

(Equation 9) ... (Equation 11)

Further, similarly to the method described in the third embodiment, it is possible to calculate the Y coordinate of the position where the boundary between the region 1504 visible and the region 1503 hidden on the image is captured.

When searching for a similar area on the image at the next time t + Δt, the template 1601 extracted at the time t is divided into two areas 1504 that are hidden and visible in the image as shown in FIG. The degree of similarity with the surrounding area searched for for each area is calculated. Then, assuming that the similarity in the region 1504 visible on the image is S1 and the similarity of the region 1503 hidden on the image is S2, S1 and S2 are weighted, and the similarity of the entire template is calculated as in Expression 12. calculate. S = α1 × S1 + α2 × S2 (Equation 12) Since an object different from the object being tracked is captured in the template in the area 1503 hidden on the image, it is not appropriate to use it for template matching. Therefore,
α2 = 0 or a value close to 0.

Although the method of dividing the template into two regions has been described in the present embodiment, the template may be divided into three or more regions. In that case, the weight of the region above the template may be increased, and the weight of the region below the template may be reduced.

As described above, according to the fifth embodiment of the present invention, the weight of the similarity obtained from the hidden part of the template is reduced, so that a plurality of objects overlap on the image, Even if is not visible, the object can be tracked with high accuracy continuously.

(Sixth Embodiment) The sixth embodiment of the present invention relates to the movement position calculation unit 107 described in the first embodiment or the long distance movement described in the third embodiment. The position calculation unit 803 has the configuration shown in FIG. In FIG. 17, reference numeral 1701 denotes an object imaging position calculation unit that calculates a position where an area of an object detected by the object detection unit 106 is imaged on an image, and 1702 denotes an imaging position calculated by the object imaging position calculation unit 1701. An edge histogram creating unit for creating an edge image and creating an edge histogram by scanning in a direction perpendicular to the moving direction of the object.
An edge histogram storage unit 1704 for storing the edge histogram created by the step S02, an edge histogram 1704 for performing matching at a certain time and the next time using the edge histogram stored in the edge histogram storage unit 1703 and calculating a moving position of the object. It is a matching unit.

Next, the operation of the above embodiment will be described. First, the object imaging position calculation unit 1701 calculates the position Yr in the Y-axis direction where the rear end position of the vehicle is imaged on the image by the method described in the first embodiment. Further, in the edge histogram creation unit 1702, Y <Yr
Then, an edge image is created and scanned in the X-axis direction to create an edge histogram 506 on the Y-axis 505 in FIG. Then, the edge histogram storage unit 1703 stores the edge histogram 506 created by the edge histogram creation unit 1702.

Next, the edge histogram matching unit 1
In 704, matching is performed using the edge histogram stored in the edge histogram storage unit 1703 to calculate the moving position of the vehicle. Matching method is 3rd
Short-range movement position calculation unit 80 described in the embodiment
It can be realized by a method similar to the method of calculating the movement position from the projection image in the operation 2. However, in edge histogram matching, the V axis of the projected image is the Y axis, and the height H of the projected image is the number of edge pixels.

As described above, according to the sixth embodiment of the present invention, an object is tracked by creating an edge histogram from an object area detected based on distance information. Thus, it is possible to reduce the amount of calculation at the time of matching and to accurately track an object.

(Seventh Embodiment) A seventh embodiment of the present invention has the configuration shown in FIG. FIG.
, 101, 102, 104 to 108 are the same as those described in the first embodiment,
05 is the same as that described in the fourth embodiment. 1801 is 3 from the image captured by the imaging unit.
A ternary edge image creation unit for creating a value edge image, 1
An associating unit 802 measures the distance information of the imaging space by associating a plurality of ternary edge images created by the ternary edge image creating unit.

Next, the operation of the above embodiment will be described. First, an edge image is created by the ternary edge image creation unit 1801 from the images captured by the imaging units 101 and 102. The associating unit 1802 measures the distance information by associating the edge images created by the ternary edge image creating unit 1801 with each other. An example of a method for measuring distance information in association with a ternary edge image is disclosed in
-294143. In this technology,
The captured image is developed into a plurality of frequency component images, and second derivative processing is performed on each of the frequency component images.
A ternary image is created by performing a binarization process. Then, a one-dimensional window of size N is set on one ternary frequency component image serving as a reference, and a region similar to the ternary pattern of N pixels in the one-dimensional window is set to the ternary frequency of the other image. A search is performed from the component image to determine a corresponding area. Further, the entire image is divided into blocks each having a size of M × N pixels, and the corresponding region of the block is determined with reference to the corresponding region of the one-dimensional window included in each block.

Distance information storage unit 104, object projection unit 10
5. Object detection unit 106, object characteristic part detection unit 1101
Performs the same operation as that described in the first embodiment or the fourth embodiment. The characteristic part imaging position detection unit 1102 calculates an area where the characteristic part of the vehicle is imaged on the image, and the template 1103 images the vehicle characteristic part on the edge image created by the ternary edge image creation unit 1801. The extracted area is extracted as a template.
After that, the template storage unit 1104, the template matching unit 1105, and the object information calculation unit 108 perform operations similar to those described in the first embodiment and the fourth embodiment. In the present embodiment, the case has been described where the ternary edge image created by the ternary edge image creating unit is used by the template extracting unit according to claim 4. Can be realized by the same operation.

As described above, according to the seventh embodiment of the present invention, matching is performed using a template extracted from a ternary edge image created when measuring distance information by stereo image processing. It is possible to reduce the amount of calculation at the time of matching and improve the matching accuracy.

(Eighth Embodiment) An eighth embodiment of the present invention has the configuration shown in FIG. FIG.
In the figure, 101 to 108 are the same as those described in the first embodiment, 1901 is an object length calculation unit for calculating the length of the area of the object detected by the object detection unit 106, and 1902 is the object length calculation An object length storage unit that stores the object length detected by the unit 1901.

Next, the operation of the above embodiment will be described. First, 101 to 106 perform the same operation as that described in the first embodiment to detect an area where an object exists. The object length calculation unit 1901 calculates the length of the object from the detected projected image of the object. Here, the length of the object can be calculated as the length of the object projection image in the V-axis direction. The object length storage unit 1902 stores the object length individually for each detected vehicle. Then, the movement position calculation unit 107 individually tracks the detected vehicles and calculates the movement position.

Next, the operation of the object information calculation unit 108 will be described with reference to FIG. At a certain time, three vehicles (vehicle A200)
1. It is assumed that the movement positions of the vehicle B2002 and the vehicle C2003) have been calculated. Then, the positions of the vehicles A, B, and C can be detected at V = V1, V2, and V3, respectively.
It is assumed that the vehicle lengths of the vehicles A, B, and C stored in 902 are L = L1, L2, and L3, respectively.

In the range of the image, Vs>
The range of V> Ve is set as the measurement range 2004 of the space occupancy. Here, the measurement range 2004 may be the entire imaging range or a part of the imaging range. At this time, Vs> V1,
If V2, V3> Ve, the space occupancy can be calculated by Expression 13. Space occupancy = (L1 + L2 + L3) / (Vs-Ve) (Equation 13)

Further, when the vehicle is within the range of Vs>V> Ve, N
If there are multiple vehicles, the space occupancy is calculated as
Can be calculated. Space occupancy = (L1 + L2 +... + LN) / (Vs-Ve) (Equation 14)

In the present embodiment, the object information calculation unit uses the length information of the object stored in the object length storage unit and the position information of the object calculated by the movement position calculation unit according to claim 1. The space occupancy of the object is calculated by using the position information of the object calculated by the long-distance range movement position calculation unit according to the third aspect.

As described above, according to the eighth embodiment of the present invention, the space occupancy is calculated by using the length information of the object measured by the stereo image processing. In addition, the space occupancy can be calculated with high accuracy even when shooting far away.

[0112]

As described above, the object tracking device according to the present invention has the following effects. By detecting an object using the distance information measured by the stereo image processing and tracking the object using a feature amount other than the distance information, it is possible to accurately track the object over a wide range.

By calculating the actual width of the object and judging whether the objects detected in two adjacent specific areas are the same vehicle, an object existing over the two specific areas can be accurately detected / tracked. it can.

In the short distance range, the object is tracked by using the distance information measured by the stereo image processing. In the long distance range, the object is tracked by using a feature amount other than the distance information. Tracking can be realized with high speed and high accuracy.

The characteristic part of the object is determined from the projected image of the object, the area where the characteristic part is imaged on the image is calculated, and the template is extracted. It is possible to reduce the amount of calculation at the time of template matching and achieve accurate object tracking.

From the position of another object existing around the object being tracked, a portion of the tracked object that is not visible due to overlapping on the template is predicted, and the template similarity weight of this portion is reduced during template matching. By doing so, the tracking of the object can be continued even if the overlap occurs.

Object tracking is performed by creating an edge histogram from the region of the object detected by stereo image processing, so that an edge histogram required for tracking can be created accurately and without excess or deficiency. Good tracking of the object can be realized.

Since matching is performed using a template or an edge histogram extracted from a ternary edge image created when measuring distance information by stereo image processing, it is not necessary to create a new edge image, and the amount of calculation at the time of tracking is performed. Can be reduced, and the template can be extracted from the ternary edge image, so that tracking can be performed with high accuracy.

By calculating the space occupancy using the length information of the object measured by the stereo image processing, the space occupancy can be calculated with high accuracy both in the case of bird's eye view shooting and in the case of shooting a long distance range.

[Brief description of the drawings]

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

FIG. 2 is a schematic diagram illustrating a relationship between a rectangular area in an image and pixels according to the first embodiment of the present invention.

FIG. 3 shows (X, Y,
K) Schematic diagram showing the relationship between the coordinate system and the (U, V, H) coordinate system

FIG. 4 is a schematic diagram showing a vehicle projection image and a vehicle area according to the first embodiment of the present invention.

FIG. 5 is an explanatory diagram showing a method of calculating an object imaging position according to the first embodiment of the present invention.

FIG. 6 is a block diagram illustrating a configuration of an object tracking device according to a second embodiment of the present invention.

FIG. 7 is a schematic diagram illustrating a vehicle traveling across two lanes according to a second embodiment of the present invention.

FIG. 8 is a block diagram illustrating a configuration of an object tracking device according to a third embodiment of the present invention.

FIG. 9 is a schematic diagram showing a relationship between a short distance range and a long distance range according to the third embodiment of the present invention.

FIG. 10 is an explanatory diagram of a projection image tracking device according to a third embodiment of the present invention.

FIG. 11 is a block diagram illustrating a configuration of an object tracking device according to a fourth embodiment of the present invention.

FIG. 12 is a schematic diagram showing a relationship between a vehicle projection image and a vehicle feature according to a fourth embodiment of the present invention.

FIG. 13 is an explanatory diagram of a method of calculating an imaging position of an object feature unit according to a fourth embodiment of the present invention.

FIG. 14 is a block diagram illustrating a configuration of an object tracking device according to a fifth embodiment of the present invention.

FIG. 15 is a schematic diagram illustrating overlapping of vehicles according to a fifth embodiment of the present invention.

FIG. 16 is a schematic diagram illustrating division of a template according to the fifth embodiment of the present invention.

FIG. 17 is a block diagram illustrating a configuration of an object tracking device according to a sixth embodiment of the present invention.

FIG. 18 is a block diagram illustrating a configuration of an object tracking device according to a seventh embodiment of the present invention.

FIG. 19 is a block diagram showing a configuration of an object tracking device according to an eighth embodiment of the present invention.

FIG. 20 is an explanatory diagram of a method of calculating a space occupancy of a vehicle according to an eighth embodiment of the present invention.

[Explanation of symbols]

 101, 102 imaging unit 103 distance measurement unit 104 distance information storage unit 105 object projection unit 106 object detection unit 107 moving position calculation unit 108 object information calculation unit 201 left image 202 pixel 203 rectangular area 301 imaging surface 302 road surface 401 projected image 402 Vehicle area 501 Road surface 502 Lane 503 X axis 504, 506 Edge histogram 505 Y axis 601 Width calculation unit 602 Same object determination unit 701, 702 Lane 703 X axis 704 Edge histogram 801 Distance determination unit 802 Short distance range movement position calculation unit 803 Long-distance range movement position calculation unit 901 Imaging range 902 Short-distance range 903 Long-distance range 1001 Projection image 1002 Rear end position 1003 Projection image at next time 1004 Rear end position at next time 1005 Partial projection image 1006 After partial projection image Position 1101 Object feature part extraction unit 1202 Feature part imaging position calculation unit 1203 Template extraction unit 1204 Template storage unit 1205 Template matching unit 1201 Vehicle feature unit 1301 Road surface 1302 Lane 1303 X axis 1304 Edge histogram 1401 Overlap determination unit 1402 Template matching unit 1403 Area matching section 1404 Similarity calculation section 1405 Moving position determination section 1501, 1502 Vehicle 1503 Hidden area 1504 Visible area 1601 Template 1701 Object imaging position calculation section 1702 Edge histogram creation section 1703 Edge histogram storage section 1704 Edge histogram matching section 1801 ternarization Edge image creation unit 1802 Correlation unit 1901 Object length calculation unit 1902 Object Long storage unit 2001 Vehicle A 2002 Vehicle B 2003 Vehicle C 2004 Measurement range

──────────────────────────────────────────────────の Continued on the front page (51) Int.Cl. ⁶ Identification symbol FI G01P 3/36 G01P 3/36 C G01S 3/786 G01S 3/786 G06T 7/00 H04N 5/232 C 7/20 G06F 15 / 62 415 H04N 5/232 15/70 410

Claims (8)

[Claims] 1. A plurality of image pickup units arranged at predetermined intervals, a distance measurement unit for measuring distance information of an image pickup space from a plurality of images picked up by the image pickup unit by stereo image processing, and a distance at a reference time. A distance information storage unit for storing information, and a plane formed by the traveling direction of the object and the perpendicular direction of the observation plane based on the distance information obtained from the distance measurement unit and the distance information at the reference time stored in the distance information storage unit. An object projection unit that projects an object image on an observation plane for each specific region in space, and an object detection unit that individually detects regions of a plurality of objects from an object projection image obtained by the object projection unit.
A moving position calculating unit that calculates a moving position of the object by performing matching at a certain time and a next time using a feature amount other than the distance information for each object detected by the object detecting unit, and the moving position calculating unit. An object tracking device, comprising: an object information calculation unit configured to calculate the number of objects passing through a predetermined position, a moving speed of the object, a size of the object, and an abnormal movement of the object based on an obtained result. 2. A width calculation unit, wherein the object detection unit individually calculates the width of a plurality of objects detected by the object detection unit;
The object tracking device according to claim 1, further comprising: an identical object determination unit that determines whether objects detected in two adjacent specific regions are the same object based on the width of the object calculated by the width calculation unit. 3. A moving position calculating section for determining whether a position of an object detected by an object detecting section is a short distance or a long distance from an imaging section, and determining distance information in a short distance range from the imaging section. A short-range moving position calculating unit that calculates a moving position of an object by performing matching at a time used as a feature amount and a next time, and a time using a feature amount other than distance information in a range far from the imaging unit. 3. The object tracking device according to claim 1, further comprising: a long-distance range movement position calculation unit that performs matching at a next time and calculates a movement position of the object. 4. An object feature portion extraction unit, wherein the movement position calculation unit or the long distance range movement position calculation unit extracts a characteristic portion of the object from the object projection image created by the object projection unit, and the object feature A characteristic part imaging position calculation unit that calculates a position where the object characteristic part extracted by the part extraction unit is captured in the image; and information other than distance information from the position calculated by the characteristic part imaging position calculation unit as a template. A template extraction unit to be extracted; a template storage unit that stores the template extracted by the template extraction unit; and a moving position of the object by performing matching at one time and the next time using the template stored in the template storage unit. 4. The object according to claim 1, further comprising a template matching unit for calculating Tracking device. 5. An overlap judging unit, wherein a template matching unit judges an overlap of objects on an image from positions of the plurality of objects detected by the object detection unit, and the plurality of objects overlap by the overlap judgment unit. A template dividing unit that divides the template into a plurality of regions when it is determined that the region is divided by the template dividing unit; a region matching unit that performs matching for each region divided by the template dividing unit; 5. The object tracking device according to claim 4, further comprising: a similarity calculating unit that calculates the similarity of the entire template; and a moving position determining unit that determines a moving position of the object based on a result calculated by the similarity calculating unit. 6. An object imaging position calculation unit, wherein a movement position calculation unit or a long-distance range movement position calculation unit calculates a position on an image of an object from a position of the object detected by the object detection unit; An edge histogram creating unit that creates an edge image for the imaging position calculated by the calculating unit and scans in a direction perpendicular to the moving direction of the object to create an edge histogram, and an edge histogram created by the edge histogram creating unit. 2. An edge histogram storage unit for storing, and an edge histogram matching unit for performing matching at a certain time and a next time using the edge histogram stored in the edge histogram storage unit to calculate a moving position of the object. The object tracking device according to claim 2. 7. A ternary edge image creating section for creating a ternary edge image from an image captured by the image capturing section, and a plurality of 3D edge images created by the ternary edge image creating section. A ternary edge image created by the ternary edge image creating unit by a template extracting unit or an edge histogram creating unit. The object tracking device according to any one of claims 1, 2, 3, 4, 5, and 6 used. 8. An object length calculation unit for calculating an object length from an object projection image created by the object projection unit,
An object length storage unit that individually stores the object length calculated for each object by the object length calculation unit, wherein the object information calculation unit includes information on the length of the object stored in the object length storage unit, The space occupancy of an object is calculated using the position information of the object calculated by the position calculation unit or the long-distance range movement position calculation unit, according to claim 1, claim 2, claim 3, claim 4, claim 5, or claim 5. The object tracking device according to claim 6 or 7.