JP7017217B1 - Target tracking device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an object tracking device for tracking an object reflected in a time-series image. A target tracking device 1 has a tracking unit 3 that tracks a plurality of objects reflected in a series of time-series images X, and a measurement unit that measures a distance between the objects in a series of time-series images X. Based on 4 and the measured distance Y between the first target A and the second target B, it is determined whether or not the second target B is approaching the first target A. The determination unit 5 and the second target B determined to be approaching in the first time-series image X1 cannot be tracked by the tracking unit 3 in the second time-series image X2, and the first time-series image When the distance Y between the objects between the first object A and the second object B in X1 is within the first predetermined distance, the tracking is performed in the time series images after the second time series image X2. It includes a determination unit 7 that determines to treat the continuing first object A as an integral part of the second object B. [Selection diagram] FIG. 4

Description

The present invention relates to an object tracking device for tracking an object reflected in a time-series image.

Conventionally, a behavior estimation device that estimates the behavior of a behavioral body based on the displacement of the coordinates and depths of a plurality of joints of the behavioral body in a time-series image has been known (see, for example, Patent Document 1).

Patent No. 6525179

In the above technique, it is necessary to identify (track) the same behavior in a time-series image, but if a plurality of behaviors overlap, the tracking of the behavior may fail.

Therefore, an object of the present invention is to provide an object tracking device capable of tracking an object reflected in a time-series image with high accuracy.

The present invention is a tracking unit that tracks a plurality of objects reflected in a series of time-series images, and a measurement that measures the distance between the objects in at least a plurality of time-series images in the series of time-series images. A determination unit for determining whether or not the second object is approaching the first object based on the distance between the unit and the measured distance between the first object and the second object. The second target, which was determined to be approaching in the first time-series image, cannot be tracked by the tracking unit in the second time-series image following the first time-series image, and When the distance between the objects in the first time-series image between the first object and the second object is within the first predetermined distance, the time series after the second time-series image. Provided is an object tracking device comprising: a determination unit that determines to treat a first object whose tracking is continued in an image as an integral part of the second object. There is.

Further, in the third time-series image following the time-series image in which the first object is treated as integrated with the second object, the tracking unit newly starts tracking the third object. Moreover, when the distance between the objects in the third time-series image between the first object and the third object is within the second predetermined distance, the determination unit determines the third object. In the time-series images after the time-series image, the treatment of the first object as one integrated with the second object is stopped, and the third object newly started to be tracked is the second object. It is preferable to decide to treat it as an object.

Further, among the tracking objects, a counting unit for counting the objects that have passed through the detection lines virtually arranged at the same spatial position in each time-series image is further provided, and is integrated with the second object. When the first object, which is treated as a ghost, passes through the detection line, it is preferable that the counting unit adds the second object to count.

Further, according to another aspect of the present invention, a program installed on a computer, in which a step of tracking a plurality of objects reflected in a series of time-series images, and at least a plurality of the series of time-series images. Based on the step of measuring the distance between objects in the time-series image of the above and the distance between the objects measured between the first object and the second object, the second object is the second object. The step of determining whether or not the object is approaching 1 and the second object determined to be approaching in the first time-series image are the second following the first time-series image. When the tracking cannot be performed in the time-series image of the above, and the distance between the objects in the first time-series image between the first object and the second object is within the first predetermined distance. In addition, in the time-series images after the second time-series image, the step of determining to treat the first object whose tracking is continued as one integrated with the second object is provided. It provides a target tracking program characterized by this.

Further, according to another aspect of the present invention, a step of tracking a plurality of objects reflected in a series of time-series images and a plurality of objects in at least a plurality of time-series images in the series of time-series images. Whether or not the second target is approaching the first target based on the step of measuring the distance between the objects and the distance between the measured first target and the second target. The step of determining whether or not the second target, which was determined to be approaching in the first time-series image, cannot be tracked in the second time-series image following the first time-series image. Moreover, when the distance between the objects in the first time-series image between the first object and the second object is within the first predetermined distance, the second time-series image and thereafter. Provided is an object tracking method comprising: a step of determining to treat a first object whose tracking is continued as one integrated with the second object in the time-series image of the above. is doing.

Further, according to another aspect of the present invention, there is a tracking unit that tracks a plurality of objects reflected in a series of time-series images, and a plurality of objects in at least a plurality of time-series images in the series of time-series images. The first target and the second target are approaching based on the distance between the measuring unit for measuring the distance between the objects and the measured distance between the first target and the second target. A determination unit for determining whether or not the image is approaching, and an object between the first object and the second object in the first time-series image, which is determined to be approaching. When the distance is equal to or less than the first predetermined distance, it is decided to treat the first object as one integrated with the second object in the time series images after the first time series image. Even if the tracking unit cannot track the second object in the time-series image after the first time-series image, the determination unit continues the tracking. Provided is an object tracking device characterized in that it maintains a decision to treat the first object as an integral part of the second object.

Also, according to another aspect of the invention, a computer-installed program that tracks a plurality of objects in a series of time-series images and at least a plurality of the series of time-series images. Based on the step of measuring the distance between objects in the time-series image of the above and the distance between the objects measured between the first object and the second object, the first object and the first object are described. The step of determining whether or not the second object is approaching, the first object and the second object in the first time-series image, which are determined to be approaching, and the first time-series image. When the distance between the objects is equal to or less than the first predetermined distance, the first object is integrated with the second object in the time-series images after the first time-series image. The step of determining the second object is provided even when the tracking unit cannot track the second object in the time-series image after the first time-series image. Provided in the subject tracking program, characterized in that the determination to treat the first subject to which the tracking is continued as an integral part with the second subject is maintained.

Further, according to another aspect of the present invention, a step of tracking a plurality of objects reflected in a series of time-series images and between the plurality of objects in at least a plurality of time-series images in the series of time-series images. Whether or not the first target and the second target are approaching based on the step of measuring the distance between the objects and the distance between the measured first target and the second target. The step of determining whether or not, and the distance between the objects determined to be approaching in the first time-series image and between the first object and the second object in the first time-series image are When the distance is equal to or less than the first predetermined distance, in the time-series images after the first time-series image, the step of deciding to treat the first object as one integrated with the second object. , And even if the tracking unit cannot track the second target in the time-series image after the first time-series image, the tracking is continued in the determination step. Provided is a subject tracking method characterized in that the determination to treat the subject as one with the second subject is maintained.

According to the object tracking device of the present invention, it is possible to track an object reflected in a time-series image with high accuracy.

Explanatory drawing of tracking by embodiment of this invention Block diagram of the target tracking device according to the embodiment of the present invention Explanatory drawing of distance between objects by embodiment of this invention Flow chart of the target tracking device according to the embodiment of the present invention Explanatory drawing of distance between objects by modification of this invention

Hereinafter, the target tracking device 1 according to the embodiment of the present invention will be described with reference to FIGS. 1 to 4.

As shown in FIG. 1, the object tracking device 1 is for tracking a plurality of objects (first object A and second object B in FIG. 1) reflected in a series of time-series images X. .. In the present embodiment, a human being will be used as an example for explanation.

As shown in FIG. 2, the target tracking device 1 includes an acquisition unit 2, a tracking unit 3, a measurement unit 4, a determination unit 5, a determination unit 6, and a counting unit 7.

The acquisition unit 2 acquires a series of time-series images X. In the present embodiment, it is assumed that a series of frames constituting an image of a predetermined space shot by a fixed shooting means is acquired as a series of time-series images X.

The tracking unit 3 tracks a plurality of objects (in the present embodiment, the first object A and the second object B) reflected in a series of time-series images X.

Various tracking methods can be considered, but in the present embodiment, the target is specified based on the aggregate of a plurality of joints detected in one time-series image X, and the aggregate of the detected joints is specified. When the aggregate corresponding to is detected in the next time-series image X, it is identified as the same target and tracked.

The measuring unit 4 measures the distance Y between objects in at least a plurality of time-series images X in a series of time-series images X. In the present embodiment, the distance Y between objects in each time-series image X is measured, but the distance Y between objects may be measured for each predetermined number of images, or the distance Y between objects may be measured only in any time-series image X. May be measured. The distance Y between the objects may be an absolute distance or a relative distance.

In this embodiment, the distance Y between objects on the horizontal plane shown in FIG. 3A is calculated from the time-series image X taken from the side shown in FIG.

Various methods can be considered for this calculation method, but in the present embodiment, the first target A and the second target A and the second target are selected from the sizes of the first target A and the second target B reflected in each time-series image X. The coordinates of B on the horizontal plane are estimated, and the distance between the coordinates is calculated as the distance Y between the objects.

The determination unit 5 determines whether or not the second target B is approaching the first target A based on the measured distance Y between the first target A and the second target B. To judge.

This may be determined as "approaching" when the inter-object distance Y in one time-series image X is shorter than the inter-object distance Y in the immediately preceding time-series image X, or may be determined to be "approaching". When the distance between the objects in the time-series image X is shorter than the distance Y, it may be determined as "approaching".

As shown in FIGS. 3 (b) and 3 (c), the determination unit 6 converts the second target B determined to be “approaching” in the first time-series image X1 into the first time-series image X1. In the subsequent second time-series image X2, the tracking unit 3 cannot be tracked, and the distance Y between the first target A and the second target B in the first time-series image X1 is the first. When it is within a predetermined distance, in the time series image X after the second time series image X2, the first target A whose tracking is continued is treated as integrated with the second target B. decide.

This means that "the second target B, which was" approaching "the first target A until then, can no longer be tracked" means that "the second target B should overlap with the first target A." It is due to the idea that "is."

This makes it possible to easily detect the overlap of objects with a low load without providing a high-performance and high-load configuration for detecting the overlap of objects, thereby realizing highly accurate tracking of objects. It becomes possible.

The timing of handling as a unit depends on the ability of the tracking unit 3. For example, if the ability of the tracking unit 3 is high, it will be treated as integrated only when most of the first target A and the second target B overlap, and if the ability of the tracking unit 3 is low, the first Even if a part of the object A and the second object B overlap each other, they can be treated as one.

Further, in the present embodiment, the first is provided that the distance Y between the objects in the first time-series image X1 between the first object A and the second object B is within the first predetermined distance. In the time-series image X after the time-series image X2 of 2, the first target A is treated as integrated with the second target B, so that the second target B is far from the first target A. Even when the second target B cannot be tracked for some reason even though it is distant, it is possible to prevent the second target B from being treated as integrated with the first target A.

As the first predetermined distance, if an appropriate distance is set after obtaining "how long the distance Y between the objects becomes so that the tracking unit 3 cannot track the second object B" by an experiment or the like. good.

Further, as shown in FIG. 3D, the determination unit 6 treats the first target A as an integrated body with the second target B, and the third time-series image X3 following the time-series image X. In, the tracking unit 3 newly starts tracking the third target C, and the distance Y between the first target A and the third target C in the third time-series image X3 is a second predetermined distance. If it is within a distance, in the time series image X after the third time series image X3, the treatment of the first target A as one integrated with the second target B is stopped, and tracking is newly performed. It is decided to treat the started third object C as the second object B.

This means that "the tracking of the third target C was newly started at a position close to the first target A" means that "the second target B that overlapped with the first target A is the first target A." It is due to the idea that "it should be away from."

This makes it possible to detect the separation of the target easily and with a low load without providing a high-performance and high-load configuration for detecting the separation of the target, thereby realizing highly accurate tracking of the target. It becomes possible.

The timing of treating the third target C as the second target B also depends on the ability of the tracking unit 3. For example, if the ability of the tracking unit 3 is high, even if the first object A and the second object B are separated to some extent, the third object C will be treated as the second object B, and the ability of the tracking unit 3 If is low, the third object C can be treated as the second object B only after the first object A and the second object B are completely separated.

Further, in the present embodiment, the third object is provided on the condition that the distance Y between the objects A and the third object C in the third time-series image X3 is within the second predetermined distance. Since the target C is treated as the second target B, the third target C is treated as the second target C even when the tracking of the third target C is started at a position far away from the first target A. It is prevented from being treated as the target B of.

As the second predetermined distance, "when the inter-target distance Y becomes, the tracking unit 3 starts to individually track the second target B that overlaps with the first target A" by an experiment or the like. After that, you can set an appropriate distance.

The counting unit 7 counts the objects being tracked that have passed through the detection lines Z (see FIGS. 1 and 3) virtually arranged at the same spatial position in each time-series image X.

Further, in the present embodiment, when the first target A, which is treated as integrated with the second target B, passes through the detection line Z, the counting unit 7 adds the second target B. And count.

This prevents the second target B, which overlaps the first target A, from failing to be counted.

Subsequently, the operation of the target tracking device 1 will be described with reference to the flowchart of FIG. In the following, the description will be given from the state in which the tracking of the target A and the target B has already been performed.

First, one time-series image X is acquired (S1), and it is determined whether or not the tracking unit 3 can track the second target B in the acquired time-series image X (S2).

If the second target B can be tracked (S2: YES), the process returns to S1 and the next time-series image X is acquired.

On the other hand, when the second target B cannot be tracked (S2: NO), it is determined as "approaching" in the previous time-series image X (first time-series image X1), and the first time-series image X1 is determined. It is determined whether or not the distance Y between the objects in the time-series image X1 is within the first predetermined distance (S3).

If it is not "approaching" or "within the first predetermined distance" (S3: NO), the process returns to S1 and the next time-series image X is acquired.

On the other hand, if it is "approaching" and "within the first predetermined distance" (S3: YES), it is decided to treat the first target A as one integrated with the second target B (S3: YES). S4).

After that, the following time-series image X is acquired (S5), and it is determined whether or not the tracking unit 3 has newly started tracking the third target C (S6).

If the tracking of the third target C has not been newly started (S6: NO), the process returns to S5 and the subsequent time-series image X is acquired.

On the other hand, when the tracking of the third target C is newly started (S6: YES), the distance Y between the first target A and the third target C in the time-series image X is the second. It is determined whether or not it is within a predetermined distance (S7).

If the distance Y between the objects is not within the second predetermined distance (S7: NO), the process returns to S5 and the next time-series image X is acquired.

On the other hand, when the distance Y between the objects is within the second predetermined distance (S7: YES), it is decided to treat the third object C as the second object B (S8), and then the process returns to S1. , Acquires the following time-series image X.

When the first target A passes the detection line Z at any timing (time-series image X), whether or not the first target A is treated as integrated with the second target B. The target is counted based on.

As described above, in the present embodiment, when the second target B, which has been "approaching" to the first target A, cannot be tracked, the first time-series image X will be the first. The target A of the above is treated as integrated with the second target B.

With such a configuration, it is possible to easily detect the overlap of the targets with a low load without providing a high-performance and high-load configuration for detecting the overlap of the targets, and thus the height of the target. It is possible to realize accurate tracking. Further, assuming that the first target A is integrated with the second target B, provided that the distance Y between the targets between the first target A and the second target B is within the first predetermined distance. As a result, the second target B is handled even when the second target B cannot be tracked for some reason even though the second target B is far away from the first target A. It is prevented from being treated as one with the object A of 1.

Further, in the present embodiment, when the tracking of the third target C is newly started in the situation where the first target A is treated as integrated with the second target B, the third target C is used. Is treated as the second object B.

With such a configuration, it is possible to detect the separation of the target easily and with a low load without providing a high-performance and high-load configuration for detecting the separation of the target, and thus the high accuracy of the target. It is possible to realize a good tracking. Further, the third target C is treated as the second target B on condition that the distance Y between the targets between the first target A and the third target C is within the second predetermined distance. , It is possible to prevent the third target C from being treated as the second target B until the tracking of the third target C is started at a position far away from the first target A.

Further, in the present embodiment, when the first target A, which is treated as being integrated with the second target B, passes through the detection line Z, the second target B is added and counted.

According to such a configuration, it is possible to prevent the second target B overlapping the first target A from failing to be counted.

The target tracking device of the present invention is not limited to the above-described embodiment, and various modifications and improvements can be made within the scope described in the claims.

For example, in the above embodiment, "when the second target B, which has been" approaching "the first target A until then, cannot be tracked, the first target A is used in the subsequent time-series image X. Was treated as integrated with the second target B ”, but the first target A may be treated as integrated with the second target B even before the second target B cannot be tracked.

In this case, as shown in FIG. 5, the determination unit 6 is determined to be "approaching" in the first time-series image X1 and with the first target A in the first time-series image X1. When the distance Y between the objects with the second object B is equal to or less than the first predetermined distance, the first object A is referred to as the second object A in the time series images X after the first time series image X1. It is decided to treat it as integrated with the target B, and even if the tracking unit 3 cannot track the second target B in the time series image X after the second time series image X2, it is tracked. Maintains the decision to treat the continuing first object A as one with the second object B ". That is, in this modification, the step S2 in FIG. 4 is not performed.

According to such a configuration, it is possible to predict that the second target B will overlap with the first target A, and treat the first target A as one with the second target B as soon as possible. It is possible to perform stable tracking. Further, since the step S2 in FIG. 4 is not performed, the load on the computer can be reduced.

Also in this case, "In the second time-series image X2 following the time-series image X in which the second target B could not be tracked, the tracking unit 3 newly starts tracking the third target C, and , When the distance Y between the first target A and the third target C in the second time-series image X2 is within the second predetermined distance, the determination unit 6 determines the second time. In the time-series image X after the series image X2, the treatment of the first target A as one integrated with the second target B is stopped, and the third target C newly started to be tracked is the second target C. When the first target A, which is treated as integrated with the second target B, passes through the detection line Z, the counting unit 7 determines to treat it as the target B. It is possible to add the target B of 2 and count.

Further, in the above-described embodiment, the tracking of the first target A and the second target B has been described as an example, but more targets may be tracked. In that case, it is decided whether or not to handle each combination of objects as a unit.

Further, in the above embodiment, the series of time-series images X is obtained by photographing a predetermined space by a fixed photographing means, but an arbitrary space is photographed by an imaging means or the like mounted on the moving means. It may be a thing.

Further, in the above embodiment, the measuring unit 4 calculates the distance Y between the objects on the horizontal plane for the sake of easy understanding, but the distance between the objects on the time-series image X is simply measured. It may be used as a distance Y.

Further, in the above embodiment, although human beings have been described as an example, animals and robots can also be used.

The present invention can also be applied to a program or method corresponding to the processing performed by the target tracking device, and a recording medium in which the program is stored. In the case of a recording medium, the program will be installed on a computer or the like. Here, the recording medium in which the program is stored may be a non-transient recording medium. As a non-transient recording medium, a CD-ROM or the like can be considered, but the non-transitory recording medium is not limited thereto.

1 Target tracking device 2 Acquisition unit 3 Tracking unit 4 Measurement unit 5 Judgment unit 6 Determination unit 7 Counting unit X Time-series image Y Distance between targets Z Detection line

Claims (12)

A tracking unit that tracks multiple objects in a series of time-series images,
A measuring unit that measures the distance between objects between a plurality of objects in at least a plurality of time-series images in the series of time-series images.
A determination unit for determining whether or not the second target is approaching the first target based on the measured distance between the first target and the second target.
The tracking unit cannot track the second object determined to be approaching in the first time-series image in the second time-series image following the first time-series image, and the first time-series image. In the time-series images after the second time-series image, when the distance between the objects in the time-series image 1 is within the first predetermined distance between the first object and the second object. , A decision unit that decides to treat the first object for which tracking is continuing as an integral part of the second object.
Equipped with
In the third time-series image following the time-series image in which the first object is treated as integrated with the second object, the tracking unit newly starts tracking of the third object, and When the distance between the objects in the third time-series image between the first object and the third object is within the second predetermined distance, the determination unit determines the third time-series. In the time-series images after the image, the treatment of the first object as one integrated with the second object is stopped, and the newly started tracking third object is regarded as the second object. An object tracking device characterized in that it decides to handle . Among the tracked objects, a counting unit for counting the objects that have passed through the detection lines virtually arranged at the same spatial position in each time-series image is further provided.
When the first object, which is treated as integrated with the second object, passes through the detection line, the counting unit adds the second object and counts. The target tracking device according to claim 1 . A program installed on your computer
Steps to track multiple objects in a series of time series images,
A step of measuring the distance between a plurality of objects in at least a plurality of time-series images in the series of time-series images, and a step of measuring the distance between the objects.
A step of determining whether or not the second object is approaching the first object based on the measured distance between the objects of the first object and the second object.
The second object determined to be approaching in the first time-series image cannot be tracked in the second time-series image following the first time-series image, and the first one. When the distance between the objects in the time-series image between the first object and the second object is within the first predetermined distance, in the time-series images after the second time-series image, the said The step of deciding to treat the first object, which is being tracked, as an integral part of the second object,
Equipped with
The tracking of the third object is newly started in the third time-series image following the time-series image in which the first object is treated as integrated with the second object, and the third object is described. When the distance between the objects in the time-series image between the first object and the third object is within the second predetermined distance, in the step of determining the determination, after the third time-series image. In the time-series image, the treatment of the first object as one integrated with the second object is stopped, and the newly started tracking third object is treated as the second object. A target tracking program characterized by making decisions . Further provided is a step of counting the objects that have passed through the detection lines virtually arranged at the same spatial position in each time-series image among the objects being tracked.
When the first object, which is treated as integrated with the second object, passes through the detection line, the counting step is characterized in that the second object is added and counted. The target tracking program according to claim 3 . The method used by computers
Steps to track multiple objects in a series of time series images,
A step of measuring the distance between a plurality of objects in at least a plurality of time-series images in the series of time-series images, and a step of measuring the distance between the objects.
A step of determining whether or not the second object is approaching the first object based on the measured distance between the objects of the first object and the second object.
The second object determined to be approaching in the first time-series image cannot be tracked in the second time-series image following the first time-series image, and the first one. When the distance between the objects in the time-series image between the first object and the second object is within the first predetermined distance, in the time-series images after the second time-series image, the said The step of deciding to treat the first object, which is being tracked, as an integral part of the second object,
Equipped with
The tracking of the third object is newly started in the third time-series image following the time-series image in which the first object is treated as integrated with the second object, and the third object is described. When the distance between the objects in the time-series image between the first object and the third object is within the second predetermined distance, in the step of determining the determination, after the third time-series image. In the time-series image, the treatment of the first object as one integrated with the second object is stopped, and the newly started tracking third object is treated as the second object. A target tracking method characterized by determining . Further provided is a step of counting the objects that have passed through the detection lines virtually arranged at the same spatial position in each time-series image among the objects being tracked.
When the first object, which is treated as integrated with the second object, passes through the detection line, the counting step is characterized in that the second object is added and counted. The target tracking method according to claim 5 . A tracking unit that tracks multiple objects in a series of time-series images,
A measuring unit that measures the distance between objects between a plurality of objects in at least a plurality of time-series images in the series of time-series images.
A determination unit that determines whether or not the first target and the second target are approaching based on the measured distance between the first target and the second target.
When it is determined in the first time-series image that the object is approaching, and the distance between the objects in the first time-series image between the first object and the second object is equal to or less than the first predetermined distance. If there is, a determination unit that decides to treat the first object as one integrated with the second object in the time series images after the first time series image.
Equipped with
Even if the tracking unit cannot track the second object in the time-series image after the first time-series image, the determination unit may use the first object for which the tracking is continued. Maintaining the decision to treat as one with the second object,
In the second time-series image following the time-series image in which the second object could not be tracked, the tracking unit newly starts tracking the third object, and the second in the second time-series image. When the distance between the objects of 1 and the object of the third object is within the second predetermined distance, the determination unit determines the second time-series image in the time-series images after the second time-series image. It is characterized in that it decides to stop treating the object 1 as one integrated with the second object and treat the newly started third object as the second object . Target tracking device. Among the tracked objects, a counting unit for counting the objects that have passed through the detection lines virtually arranged at the same spatial position in each time-series image is further provided.
When the first object, which is treated as integrated with the second object, passes through the detection line, the counting unit adds the second object and counts. The target tracking device according to claim 7 . A program installed on your computer
Steps to track multiple objects in a series of time series images,
A step of measuring the distance between a plurality of objects in at least a plurality of time-series images in the series of time-series images, and a step of measuring the distance between the objects.
A step of determining whether or not the first object and the second object are approaching based on the measured distance between the first object and the second object.
When it is determined in the first time-series image that the object is approaching, and the distance between the objects in the first time-series image between the first object and the second object is equal to or less than the first predetermined distance. If so, in the time-series images after the first time-series image, the step of deciding to treat the first object as one integrated with the second object, and
Equipped with
Even if the tracking unit cannot track the second object in the time-series image after the first time-series image, the first object whose tracking is continued in the determination step is described. Maintaining the decision to treat as one with the second object,
The tracking of the third object is newly started in the second time-series image following the time-series image in which the second object could not be tracked, and the first object in the second time-series image is used. When the distance between the objects with the third object is within the second predetermined distance, in the step of determining the first object in the time series images after the second time series image. Is discontinued as being integrated with the second target, and the target tracking program is determined to treat the newly started third target as the second target. .. Further provided is a step of counting the objects that have passed through the detection lines virtually arranged at the same spatial position in each time-series image among the objects being tracked.
When the first object, which is treated as integrated with the second object, passes through the detection line, the counting step is characterized in that the second object is added and counted. The target tracking program according to claim 9 . The method used by computers
Steps to track multiple objects in a series of time series images,
A step of measuring the distance between a plurality of objects in at least a plurality of time-series images in the series of time-series images, and a step of measuring the distance between the objects.
A step of determining whether or not the first object and the second object are approaching based on the measured distance between the first object and the second object.
When it is determined in the first time-series image that the object is approaching, and the distance between the objects in the first time-series image between the first object and the second object is equal to or less than the first predetermined distance. If so, in the time-series images after the first time-series image, the step of deciding to treat the first object as one integrated with the second object, and
Equipped with
Even if the tracking unit cannot track the second object in the time-series image after the first time-series image, the first object whose tracking is continued in the determination step is described. Maintaining the decision to treat as one with the second object,
The tracking of the third object is newly started in the second time-series image following the time-series image in which the second object could not be tracked, and the first object in the second time-series image is used. When the distance between the objects with the third object is within the second predetermined distance, in the step of determining the first object in the time series images after the second time series image. Is discontinued as being integrated with the second target, and the target tracking method is determined to treat the newly started third target as the second target. .. Further provided is a step of counting the objects that have passed through the detection lines virtually arranged at the same spatial position in each time-series image among the objects being tracked.
When the first object, which is treated as integrated with the second object, passes through the detection line, the counting step is characterized in that the second object is added and counted. The target tracking method according to claim 11 .

Images