JP2021117760A - Object tracking device and object tracking method - Google Patents

Abstract

To improve the performance of object tracking according to an object tracking status upon the object tracking.SOLUTION: An object tracking device includes: an object detection unit 321 which detects the position of an object to be detected and included in an image taken by an image-taking device 1; a tracking unit 322 which tracks, as a tracking object, an object detected in the past by the object detection unit 321; an association unit 323 which associates the object detected by the object detection unit 321 with the tracking object tracked by the tracking unit 322; a calculation unit 325 which increases or decreases the degree of presence certainty indicating the certainty of the presence of the tracking object based on whether or not the detected object has been associated with the tracking object; and an output unit 326 which outputs an output position based on the position of the detected object in association with identification information for identifying the tracking object when the presence certainty is greater than or equal to a first threshold value.SELECTED DRAWING: Figure 2

Description

The present invention relates to an object tracking device and an object tracking method.

Conventionally, an object existing in front of a vehicle has been detected. For example, Patent Document 1 discloses a technique of detecting a vehicle traveling in front of a vehicle as a detection target object existing in front of the vehicle based on an image captured by an in-vehicle camera and tracking the detected vehicle. It is disclosed.

Japanese Unexamined Patent Publication No. 2018-18393

In the invention described in Patent Document 1, the object detected by the matching process using the template image is tracked. In the invention described in Patent Document 1, there is a problem that the object to be tracked cannot be detected unless the object is similar to the template image, and the tracking of the object cannot be performed appropriately.

The present invention has been made in view of this point, and provides an object tracking device and an object tracking method capable of improving the performance of tracking an object according to the tracking status of the object when tracking the object. The purpose is.

The object tracking device according to the first aspect of the present invention includes an object detection unit that detects the position of a detection object that is an object to be detected included in the captured image captured by the image pickup device, and the object detection unit that has detected in the past. A tracking unit that tracks an object as a tracking object, an association unit that associates the detection object detected by the object detection unit with the tracking object that the tracking unit is tracking, and the detection object that associates with the tracking object. A calculation unit that increases or decreases the existence certainty indicating the certainty that the tracking object exists based on whether or not the tracking object is present, and identifies the tracking object when the existence certainty is equal to or higher than the first threshold value. It includes an output unit that outputs an output position based on the position of the detected object in association with the identification information.

The tracking unit may end the tracking of the tracking object when the certainty of existence of the tracking object is equal to or less than the second threshold value smaller than the first threshold value.

When the distance between the acquisition unit that acquires the distance from the imaging device to the detection object and the detection object is the first distance, the distance to the detection object is larger than the first distance. A setting unit for setting the first threshold value relatively smaller than in the case may be further provided.

The object tracking device is from the imaging device to the detection object when the acquisition unit that acquires the distance from the image pickup device to the detection object and the association unit do not associate the detection object with the tracking object. As the distance is shorter, the tracking unit may further include a setting unit that sets a relatively large initial value of the existence certainty included in the new tracking object information for the tracking unit to track.

The calculation unit determines the existence of the tracking object when the current position of the tracking object is within a predetermined area including the edge of the captured image and the detection object is not associated with the tracking object. The degree may be set to be equal to or less than the second threshold value.

The tracking unit estimates the current position of the tracking object based on the state information including the estimated position immediately before the tracking object and the speed of the tracking object, and the output unit detects the detection on the tracking object. When the objects are not associated and the certainty of existence is equal to or higher than the first threshold value, the current position of the tracking object estimated by the tracking unit may be output as the output position.

The tracking unit estimates the current position of the tracking object based on the history information indicating the history of the position of the detection object associated with the tracking object, and the output unit causes the detection object to be attached to the tracking object. When the existence certainty is not related and the existence certainty is equal to or higher than the first threshold value, the current position of the tracking object estimated by the tracking unit may be output as the output position.

The object tracking method according to the second aspect of the present invention includes a step of detecting the position of a detection object, which is an object to be detected included in the captured image captured by the imaging device, and tracking an object detected in the past as a tracking object. Indicates the certainty that the tracking object exists based on the step, the step of associating the detection object with the tracking object being tracked, and whether the detection object is associated with the tracking object. A step of increasing or decreasing the existence certainty, and a step of outputting an output position based on the position of the detected object in association with the identification information that identifies the tracking object when the existence certainty is equal to or higher than the first threshold value. To be equipped with.

According to the present invention, when tracking an object, it is possible to improve the performance of tracking the object according to the tracking status of the object.

It is a figure for demonstrating the outline of the object tracking apparatus of an embodiment. It is a figure which shows the structure of the vehicle which mounted the object tracking device. It is a figure for demonstrating the calculation method of the existence certainty degree by a calculation part. It is a figure which shows another example of the output of an output position by an output part. It is a figure which shows another example of the output of an output position by an output part. It is a flowchart which shows the processing procedure of the tracking object by the object tracking device.

[Overview of object tracking device]
1A and 1B are diagrams for explaining an outline of the object tracking device of the present embodiment. FIG. 1A shows a captured image of the front of a vehicle equipped with an object tracking device. The object tracking device is realized by, for example, an ECU (Electronic Control Unit). The object tracking device detects the position of the object included in the captured image captured in front of the vehicle. In the example of FIG. 1A, the object is another vehicle traveling in front of a vehicle equipped with an object tracking device.

The object tracking device tracks the detected object as a tracking object. The object tracking device estimates the current position of the tracking object based on the history of the detected object up to a predetermined time ago. In FIG. 1A, a plurality of tracking objects tracked by the object tracking device are shown by a rectangular frame. FIG. 1A shows tracking objects of the identification information “1” to “4”, respectively. The object tracking device calculates an existence certainty that indicates the certainty that the tracking object exists. Assuming that the tracking objects of the identification information "1" to "4" are generated at the time of FIG. 1 (a), the tracking of the identification information "1" to "4" is as shown in the box of FIG. 1 (a). The existence certainty of the objects is 0.5, which is the initial value.

FIG. 1B shows an captured image captured at the next imaging timing of the captured image of FIG. 1A. The object tracking device detects the position of the detected object included in the captured image shown in FIG. 1 (b). FIG. 1B shows, in addition to the tracking objects of the identification information “1” to “4”, the tracking objects of the identification information “5” corresponding to the detection objects newly detected by the object tracking device.

The object tracking device associates the detected object with the tracking object. The object tracking device increases or decreases the certainty of existence of the tracking object based on whether the detected object is associated with the tracking object. In the example of FIG. 1B, the object tracking device associates the tracking objects with the identification information “1” to “4” with the detected detection objects. The object tracking device increases the existence certainty of the tracking objects of the identification information "1" to "4" associated with the detection object by 0.5 to make the existence certainty 1.0. On the other hand, the object tracking device sets the existence certainty of the tracking object of the identification information "5" corresponding to the newly detected detection object to 0.5, which is the initial value.

The object tracking device outputs an output position based on the detected object in association with the identification information when the existence certainty of the tracking object is larger than the first threshold value. In the example of FIG. 1B, the first threshold value is 1.0. The object tracking device associates the output positions of the tracking objects with the identification information "1" to "4" having the existence certainty of 1.0 or more with the identification information and outputs them to the automatic driving device. Although the details will be described later, the object tracking device shall calculate the velocity or acceleration, which is a differential value of the position, in addition to the output position, based on the position of the detected object and the history of the position of the tracking object. The velocity may be the relative velocity of the tracking object to the vehicle equipped with the object tracking device, or may be the absolute velocity. The acceleration may be the relative acceleration of the tracking object to the vehicle equipped with the object tracking device, or may be the absolute acceleration.

The object tracking device increases or decreases the certainty of existence of the tracking object based on whether the detected object is associated with the tracking object. The object tracking device calculates the existence certainty so that the existence certainty of the repeatedly detected tracking object becomes relatively large, so that the object is repeatedly detected even if the object cannot be detected temporarily. It is possible to suppress the output of the output position of the tracking object from being interrupted. In addition, if the existence certainty is not sufficiently high for an object that is falsely detected for a moment, the output of the output position of the tracking object can be suppressed. Therefore, the object tracking device can improve the performance of tracking an object when tracking the object.

[Vehicle configuration]
FIG. 2 is a diagram showing a configuration of a vehicle 100 equipped with an object tracking device 3. The vehicle 100 includes an image pickup device 1, an automatic driving device 2, and an object tracking device 3. The object tracking device 3 includes a storage unit 31 and a control unit 32. The image pickup device 1 takes an image of the front of the vehicle 100. The image pickup device 1 is, for example, a stereo camera including a plurality of cameras, and measures the distance from the image pickup device 1 to the detection object based on the parallax between the plurality of cameras. The image pickup apparatus 1 images the front of the vehicle 100 at a predetermined frame rate. The automatic driving device 2 is realized by, for example, the same or different ECU as the object tracking device 3. The automatic driving device 2 automatically controls the running of the vehicle 100. The automatic driving device 2 controls the running of the vehicle 100 so that there is no risk of collision with the tracking object output by the object tracking device 3.

The storage unit 31 is composed of, for example, a ROM (Read Only Memory), a RAM (Random Access Memory), or the like. The storage unit 31 stores various programs and various data for operating the control unit 32. The control unit 32 functions as an object detection unit 321, a tracking unit 322, an association unit 323, an acquisition unit 324, a calculation unit 325, an output unit 326, and a setting unit 327 by executing a program stored in the storage unit 31. ..

The object detection unit 321 detects the position of the detection object, which is the object to be detected included in the captured image captured by the image pickup device 1. Specifically, the object detection unit 321 uses the input of the trained machine learning model as an captured image, and outputs the position and type of the detected object included in the captured image to the machine learning model. The types of detection objects are, for example, large four-wheeled vehicles, medium-sized four-wheeled vehicles, ordinary four-wheeled vehicles, motorcycles, bicycles, pedestrians, and the like.

[Object Tracking]
The tracking unit 322 tracks an object detected in the past by the object detection unit 321 as a tracking object. The storage unit 31 stores identification information of the tracking object in addition to the state information including the state such as the estimated position and the speed immediately before the tracking object being tracked. The storage unit 31 stores the identification information of the tracking object being tracked, the position of the detection object detected in the past by the object detection unit 321 and the history information indicating the history associated with the detection time when the detection object is detected. May be remembered.

The tracking unit 322 reads from the storage unit 31 the estimated position, speed, etc. immediately before the tracking object included in the state information of the tracking object being tracked. The tracking unit 322 estimates the current position of the tracking object by a known tracking process based on the state such as the estimated position and speed immediately before the read tracking object. The tracking process is realized by, for example, a Kalman filter or a particle filter.

Further, the tracking unit 322 may refer to the history information and read the position of the past detection object associated with the identification information of the tracking object and the detection time up to a predetermined time from the storage unit 31. The tracking unit 322 may estimate the current position of the tracking object by a known tracking process based on the position of the past detection object of the tracked object being read and the corresponding detection time.

The tracking unit 322 reads out the tracking object information stored in the storage unit 31. The tracking object information includes identification information that identifies the tracking object, information that indicates the state and characteristics of the tracking object, and information that indicates the certainty of existence of the tracking object. The tracking unit 322 determines whether or not the existence certainty of the tracking object included in the tracking object information is greater than the second threshold value. The second threshold value shall be smaller than the first threshold value. The tracking unit 322 ends the tracking of the tracking object when the certainty of existence of the tracking object is equal to or less than the second threshold value. The tracking unit 322 deletes the tracking object information for which tracking has been completed from the storage unit 31.

[Association of discovery object with tracking object]
The association unit 323 associates the detection object detected by the object detection unit 321 with the tracking object tracked by the tracking unit 322. The association unit 323 reads the tracking object information indicating the tracking object from the storage unit 31. The state and characteristics of the tracking object included in the tracking object information are created by the association unit 323 extracting from the past detection objects associated with the identification information of the tracking object in the history information.

The association unit 323 determines the degree of difference indicating the degree of difference between the state such as the position and the edge of the detected object detected by the object detection unit 321 and the feature such as the edge and the state and the feature of the tracking object included in the read tracking object information. Ask. For example, the association unit 323 obtains the degree of difference in consideration of the distance between the position of the detection object detected by the object detection unit 321 and the current position of the tracking object estimated by the tracking unit 322. In that case, the association unit 323 may correct the degree of difference so that the greater the distance between the position of the detection object and the current position of the tracking object estimated by the tracking unit 322, the greater the degree of difference. This distance is a three-dimensional distance between the detection object and the tracking object, but may be the number of pixels on the captured image.

The association unit 323 generates a plurality of combinations in which the detection object and the tracking object are associated with each other between the plurality of detection objects detected by the object detection unit 321 and the plurality of tracking objects tracked by the tracking unit 322. .. The association unit 323 determines a combination of objects that associate the tracking object and the detection object so that the sum of the obtained differences is minimized. At that time, the detection object whose difference between each tracking object is equal to or more than a predetermined upper limit value is stored as a new tracking object. Further, a tracking object whose difference between the detected objects and the detection objects does not fall below a predetermined upper limit value is stored as a tracking object that cannot be associated at that time.

[Hungarian algorithm]
For example, the association unit 323 can uniquely determine the combination that minimizes the difference between the plurality of detection objects and the plurality of tracking objects by using the Hungarian algorithm.

The association unit 323 associates the identification information of the tracking object with the position of the detection object and the detection time when the detection object is detected, and adds the identification information to the history information stored in the storage unit 31. The association unit 323 creates new tracking object information for the tracking unit 322 to track the detection object when the detection object is not associated with the tracking object. The association unit 323 stores the created tracking object information in the storage unit 31. The association unit 323 creates identification information of the tracking object, information indicating the state and characteristics of the tracking object, and information indicating the certainty of existence of the tracking object, and includes the information indicating the existence certainty of the tracking object in the tracking object information. The association unit 323 prevents the identification information used for tracking other tracking objects from being reused for a certain period of time. The association unit 323 sets the existence certainty of the tracking object as an initial value.

The acquisition unit 324 acquires the distance from the image pickup apparatus 1 to the detection object. When the image pickup device 1 is a stereo camera, the acquisition unit 324 acquires the measurement result obtained by measuring the distance from the image pickup device 1 to the detection object by the image pickup device 1. The acquisition unit 324 may acquire a measurement result obtained by measuring the distance from the image pickup device 1 to the detection object by another distance measuring device such as a laser radar (not shown).

[Calculation of existence certainty]
The calculation unit 325 increases or decreases the certainty of existence indicating the certainty that the tracking object exists, based on whether or not the tracking object and the detection object are associated with each other. FIG. 3 is a diagram for explaining a method of calculating the existence certainty by the calculation unit 325. FIG. 3 shows the change in the certainty of existence calculated by the calculation unit 325 for a single tracking object from time k to time k + 7. It is assumed that the association unit 323 has created new tracking object information for tracking the detected object at time k. As an example in FIG. 3, the calculation unit 325 assigns an initial value of 0.5 to the existence certainty of the tracking object included in the tracking object information.

When the association unit 323 associates the detection object with the tracking object at time k + 1, the calculation unit 325 adds an increment value of 0.5 to the existence certainty of the tracking object to set the existence certainty to 1.0. When the association unit 323 associates the detection object with the tracking object at the time k + 2 and the time k + 3, the calculation unit 325 adds an increment value of 0.5 to the existence certainty of the tracking object to increase the existence certainty of 1.5 in order. And 2.0.

Further, the calculation unit 325 may increase the increment value relatively as the degree of difference when the acquisition unit 324 associates the detection object with the tracking object is smaller. For example, the calculation unit 325 adds an increment value of 0.2 to the existence certainty of the tracking object when the difference degree when the acquisition unit 324 associates the detection object with the tracking object is 10 points. On the other hand, the calculation unit 325 may add an increment value of 0.4 to the existence certainty of the tracking object when the difference degree when the acquisition unit 324 associates the detection object and the tracking object is 5 points. ..

When the association unit 323 does not associate the detection object with the tracking object at time k + 4, the calculation unit 325 subtracts the decrement value 0.5 from the existence certainty of the tracking object to set the existence certainty to 1.5. When the association unit 323 does not associate the detection object with the tracking object at time k + 5 to time k + 7, the calculation unit 325 subtracts the decrement value 0.5 from the existence certainty of the tracking object to determine the existence certainty in order. It is set to 0, 0.5 and 0.0. In the example of FIG. 3, the second threshold value for determining whether or not to end the tracking of the tracking object is 0.0, and the existence certainty of the tracking object becomes equal to or less than the second threshold value at time k + 7. The tracking by the tracking unit 322 of the tracking object is terminated.

If the association unit 323 does not associate the detection object with the tracking object located at the end of the captured image, the tracking object is likely to have moved out of the imaging range of the imaging device 1. When the current position of the tracking object is within a predetermined area including the edge of the captured image, the calculation unit 325 estimates that the tracking object has moved out of the imaging range if the tracking object is not associated with the detection object. Set the certainty of existence of the tracking object to the second threshold or less. For example, assuming that the maximum value that the existence certainty of the tracking object can take is 2.0, the calculation unit 325 subtracts the existence certainty of the tracking object by the same value as this maximum value to reduce the existence certainty. Make it below the threshold. The predetermined area is defined by those skilled in the art as, for example, a region in which the tracking object is expected to move out of the imaging range in a time shorter than the time corresponding to one frame of the imaging device 1.

The calculation unit 325 calculates the output position based on the position of the detected object. As an example, the calculation unit 325 calculates the output position by the Kalman filter using both the estimated position of the object being tracked and the position of the detected object detected by the object detection unit 321. Specifically, the Kalman filter executes the position correction step by multiplying the residual between the position of the observed object and the position of the predicted object based on the transition equation by the Kalman gain and adding it to the predicted object position.

Further, the calculation unit 325 may change the value of the setting parameter in the tracking method such as the Kalman filter based on the existence certainty of the tracking object. For example, the calculation unit 325 may reduce the value of the process noise in the transition equation of the Kalman filter as the certainty of existence of the tracking object increases. Further, the calculation unit 325 is not limited to the example of calculating the output position by the Kalman filter. For example, the calculation unit 325 may calculate the output position by the particle filter, or may simply use the position of the detected object detected by the object detection unit 321 as the output position without being based on the state space model.

The calculation unit 325 refers to the history information stored in the storage unit 31 and calculates the time differential value of the change in the position of the plurality of detection objects associated with the identification information of the tracking object as the velocity of the tracking object. The calculation unit 325 calculates the acceleration of the tracking object, which is the time derivative of the velocity of the tracking object.

As described above, the calculation unit 325 calculates the output position using both the position of the object in consideration of the positions of the plurality of past detection objects and the position of the detection object detected by the object detection unit 321. Therefore, the calculation unit 325 can prevent the output position from being erroneously output when the object detection unit 321 detects a position that is significantly different from the original position as the position of the detection object.

[Output of output position]
The output unit 326 outputs the output position calculated by the calculation unit 325 in association with the identification information that identifies the tracking object when the existence certainty of the tracking object is equal to or higher than the first threshold value. In the example of FIG. 3, the first threshold value is 1.0. Since the existence certainty of the tracking object is 1.0 at time k + 1, the output unit 326 determines that the existence certainty of the tracking object is equal to or higher than the first threshold value at time k + 1, and associates it with the identification information of this tracking object. Then, the output position is output to the automatic operation device 2. Similarly, the output unit 326 determines that the existence certainty of the tracking object is equal to or higher than the first threshold value at the time k + 2 and the time k + 3, and outputs the output position to the automatic driving device 2 in association with the identification information of the tracking object. do.

The output unit 326 determines that the existence certainty of the tracking object is equal to or higher than the first threshold value at the time k + 4 and the time k + 5. At this time, since the detection object is not associated with the tracking object by the association unit 323, the output unit 326 outputs the current position of the tracking object estimated by the tracking unit 322 as the output position. On the other hand, the output unit 326 determines that the certainty of existence of the tracking object is less than the first threshold value at the time k, the time k + 6 and the time k + 7, and does not output the output position to the automatic operation device 2.

4 and 5 are diagrams showing another example of output at the output position by the output unit 326. Since the time k to the time k + 6 in FIG. 4 are the same as those in FIG. 3, the description thereof will be omitted. The output unit 326 does not output the output position to the automatic operation device 2 because the certainty of existence of the tracking object is less than the first threshold value at the time k + 6. At this time, since the existence certainty is greater than 0.0, which is the second threshold value, the tracking of the tracking object by the tracking unit 322 is not completed.

At time k + 7, the calculation unit 325 adds the existence certainty by 0.5, which is an increment value, to set the existence certainty to 1.0, and the existence certainty becomes equal to or higher than the first threshold value. Therefore, the output unit 326 resumes the output at the output position in association with the same identification information as before the output was interrupted. By outputting the output position in association with the same identification information as before the output is interrupted, the output unit 326 can prevent the identification information of the tracking object from being unintentionally reattached.

It is assumed that a detection object that does not originally exist is erroneously detected at time k in FIG. According to the output unit 326, when the association unit 323 creates the tracking object information corresponding to the detected object at time k, the initial value 0.5 of the existence certainty of the tracking object is smaller than the first threshold value of 1.0. Judge and do not output the output position. If the association unit 323 does not associate the detection object with the tracking object at the next time k + 1, the existence certainty of the tracking object is subtracted by 0.5, which is the decrement value, to become 0.0. The output unit 326 determines that the existence certainty is smaller than the first threshold value, and does not output the output position. At this time, since the existence certainty of the tracking object is 0.0 or less, which is the second threshold value, the tracking of the tracking object by the tracking unit 322 is completed.

As described above, if an object that does not originally exist is temporarily falsely detected, it will not be associated with the detected object at a later time, so there is a possibility that the existence certainty of the tracking object will exceed the first threshold value. small. Therefore, the output unit 326 can suppress the output position of the erroneously detected object to the automatic operation device 2. Further, the output unit 326 is not limited to the example of outputting the output position of the tracking object to the automatic driving device 2. For example, the output unit 326 may output the output position of the tracking object to the display in association with the identification information. Further, in the present embodiment, the output information is described as the position of the object, but the present invention is not limited to this. The output unit 326 may output information on speed, acceleration, or magnitude, which is information other than the position defined as the state of the tracking object, to the automatic driving device 2 in association with the identification information of the tracking object.

[Variable settings]
The setting unit 327 of FIG. 2 sets various variables such as an initial value and a first threshold value. When the distance to the detection object is relatively small, it is preferable to set the first threshold value so that the output unit 326 can easily output the output position in order to suppress the contact between the vehicle 100 and the detection object.

On the other hand, when the distance to the detection object is relatively large, the output unit 326 outputs an incorrect output position by setting the first threshold value so that the output unit 326 does not easily output the output position. It can be suppressed, and the effect of suppressing the malfunction of the automatic operation by the automatic operation device 2 can be expected.

Therefore, assuming that the second distance is larger than the first distance, the setting unit 327 sets the distance to the detection object as the second distance when the distance to the detection object acquired by the acquisition unit 324 is the first distance. The first threshold value is set relatively small as compared with the case of. For example, the setting unit 327 sets the first threshold value to 1.0, which is relatively large, when the distance to the detection object is 100 meters. When the distance to the detection object is 10 meters, the setting unit 327 sets the first threshold value to 0.1, which is relatively small as compared with the case where the distance to the detection object is 100 meters. In this way, the setting unit 327 can prevent the output unit 326 from outputting an erroneous output position while ensuring the safety of the vehicle 100.

Further, the setting unit 327 may set an initial value of the existence certainty included in the new tracking object information instead of the first threshold value. The setting unit 327 may set the initial value of the existence certainty to be relatively large as the distance from the image pickup apparatus 1 to the detection object is shorter. For example, when the distance from the image pickup apparatus 1 to the detection object is 100 meters, the setting unit 327 sets the initial value of the existence certainty to 0.1, which is relatively small. On the other hand, when the distance from the image pickup apparatus 1 to the detection object is 10 meters, the initial value of the existence certainty may be set to 1.0, which is relatively large. In this way, when the distance from the image pickup apparatus 1 to the detection object is large, the setting unit 327 makes it difficult for the output unit 326 to output the output position, so that the output unit 326 outputs an erroneous output position. Can be suppressed.

Further, the setting unit 327 may set the increment value to be larger than the decrement value. By setting the increment value larger than the decrement value, the setting unit 327 is faster when the existence certainty of the tracking object becomes less than the first threshold value and the output unit 326 stops outputting the output position of the tracking object. At the timing, the output unit 326 can restart the output of the output position of the tracking object.

[Procedure for tracking tracking objects]
FIG. 6 is a flowchart showing a processing procedure for tracking a tracking object by the object tracking device 3. This processing procedure starts during the tracking of the tracking object by the tracking unit 322. When the tracking unit 322 is tracking a plurality of objects, it is assumed that the series of processes shown in FIG. 6 is repeated for each tracking object.

First, the object detection unit 321 detects the position of the detection object, which is the object to be detected included in the captured image captured by the image pickup device 1 (S101). The association unit 323 determines whether or not to associate the detection object with the tracking object (S102). When the association unit 323 associates the detection object with the tracking object in the determination of S102 (YES in S102), the calculation unit 325 adds the existence certainty of the tracking object by the increment value (S103).

The output unit 326 determines whether or not the existence certainty of the tracking object is equal to or higher than the first threshold value (S104). When the existence certainty of the tracking object is equal to or higher than the first threshold value in the determination of S104 (YES in S104), the output unit 326 automatically sets the output position of the tracking object in association with the identification information that identifies the tracking object. Output to 2 (S105). The tracking unit 322 determines whether or not the existence certainty of the tracking object included in the tracking object information is greater than the second threshold value (S106). The tracking unit 322 ends the tracking of the tracking object when the existence certainty of the tracking object included in the tracking object information is equal to or less than the second threshold value in the determination of S106 (NO in S106) (S107).

When the association unit 323 does not associate the detection object with the tracking object in the determination of S102 (NO in S102), the calculation unit 325 subtracts the existence certainty of the tracking object by the decrement value (S108), and determines in S104. Proceed to. When the existence certainty of the tracking object is less than the first threshold value in the determination of S104 (NO in S104), the output unit 326 does not output the output position of the tracking object to the automatic driving device 2 and proceeds to the determination of S106. .. The tracking unit 322 ends the process when the existence certainty of the tracking object included in the tracking object information is greater than the second threshold value in the determination of S106 (YES in S106).

[Effect of the present invention]
According to the present embodiment, the calculation unit 325 increases or decreases the certainty of existence of the tracking object based on whether or not the detection object is associated with the tracking object. The calculation unit 325 calculates the existence certainty so that the existence certainty of the repeatedly detected tracking object becomes relatively large, so that when the object cannot be detected temporarily, the repeatedly detected tracking object It is possible to suppress the interruption of the output at the output position. Therefore, the calculation unit 325 can prevent the tracking object from being unintentionally reattached with different identification information. In addition, if the existence certainty is not sufficiently high for an object that is falsely detected for a moment, the output of the output position of the tracking object can be suppressed. Therefore, the object tracking device can improve the performance of tracking an object when tracking the object.

Although the present invention has been described above using the embodiments, the technical scope of the present invention is not limited to the scope described in the above embodiments, and various modifications and changes can be made within the scope of the gist thereof. be. For example, all or a part of the device can be functionally or physically distributed / integrated in any unit. Also included in the embodiments of the present invention are new embodiments resulting from any combination of the plurality of embodiments. The effect of the new embodiment produced by the combination also has the effect of the original embodiment.

1 Imaging device 2 Automatic driving device 3 Object tracking device 31 Storage unit 32 Control unit 100 Vehicle 321 Object detection unit 322 Tracking unit 323 Association unit 324 Acquisition unit 325 Calculation unit 326 Output unit 327 Setting unit

Claims (8)

An object detection unit that detects the position of a detection object, which is an object to be detected included in the captured image captured by the image pickup device, and an object detection unit.
A tracking unit that tracks an object detected in the past by the object detection unit as a tracking object,
An association unit that associates the detection object detected by the object detection unit with the tracking object tracked by the tracking unit.
A calculation unit that increases or decreases the certainty of existence indicating the certainty that the tracking object exists based on whether or not the detection object is associated with the tracking object.
An output unit that outputs an output position based on the position of the detected object in association with the identification information that identifies the tracking object when the existence certainty is equal to or higher than the first threshold value.
An object tracking device equipped with. The tracking unit ends tracking of the tracking object when the certainty of existence of the tracking object is equal to or less than a second threshold value smaller than the first threshold value.
The object tracking device according to claim 1. An acquisition unit that acquires the distance from the imaging device to the detection object, and
When the distance to the detection object is the first distance, the setting unit which sets the first threshold value relatively smaller than the case where the distance to the detection object is the second distance larger than the first distance. And further prepare,
The object tracking device according to claim 1 or 2. An acquisition unit that acquires the distance from the imaging device to the detection object, and
When the association unit does not associate the detection object with the tracking object, the shorter the distance from the imaging device to the detection object, the more new tracking object information for the tracking unit to track the detection object. A setting unit for setting a relatively large initial value of the included certainty of existence is further provided.
The object tracking device according to claim 1 or 2. The calculation unit determines the existence of the tracking object when the current position of the tracking object is within a predetermined area including the edge of the captured image and the detection object is not associated with the tracking object. To make the degree less than or equal to the second threshold value,
The object tracking device according to claim 2. The tracking unit estimates the current position of the tracking object based on the state information including the estimated position immediately before the tracking object and the velocity of the tracking object.
When the detection object is not associated with the tracking object and the certainty of existence is equal to or higher than the first threshold value, the output unit determines the current position of the tracking object estimated by the tracking unit as the output position. Output as,
The object tracking device according to any one of claims 1 to 5. The tracking unit estimates the current position of the tracking object based on the history information indicating the history of the position of the detection object associated with the tracking object.
When the detection object is not associated with the tracking object and the certainty of existence is equal to or higher than the first threshold value, the output unit determines the current position of the tracking object estimated by the tracking unit as the output position. Output as,
The object tracking device according to any one of claims 1 to 5. The step of detecting the position of the detection object, which is the object to be detected included in the captured image captured by the image pickup device, and
Steps to track previously detected objects as tracking objects,
A step of associating the detected object with the tracking object being tracked,
A step of increasing or decreasing the certainty of existence, which indicates the certainty that the tracking object exists, based on whether or not the detection object is associated with the tracking object.
A step of outputting an output position based on the position of the detected object in association with the identification information that identifies the tracking object when the existence certainty is equal to or higher than the first threshold value.
An object tracking method that includes.

Images