CN-116647816-B - Group target tracking method by cooperation of multiple tracking devices

CN116647816BCN 116647816 BCN116647816 BCN 116647816BCN-116647816-B

Abstract

The invention provides a multi-tracking device collaborative group target tracking method which comprises the following steps of guiding tracking devices to an initial position to wait for targets to enter a view field, marking the targets according to target information, self-adaptive task allocation, allocating all the targets in the view field to the tracking devices, guiding the tracking devices to complete tracking of the targets in a multi-tracking device collaborative working mode, and performing new task allocation on the tracking devices participating in a current task after a single task is completed. The invention solves the problems in two aspects in the group target tracking process, namely the problem of group target task planning and the problem of equipment collaborative work mode design. Task planning and multi-device collaborative algorithm aiming at group targets improve the working capacity of the system.

Inventors

DENG SHUWEI
WANG ZHIWEN
SUN HANG
ZHOU SHUYU
LI SHENGQIAN
WANG YUANKANG
FENG ZHILIANG
ZHANG YAHUI
MAO YAO

Assignees

中国科学院光电技术研究所

Dates

Publication Date: 20260512
Application Date: 20230529

Claims (6)

1. A multi-tracking device collaborative group target tracking method, comprising the steps of: (1) Initial guiding, wherein each tracking device guides the tracking device to an initial position according to initial guiding information in an initial stage of starting a task to wait for a target to enter a field of view of a detector of the tracking device; (2) When multi-target detection is carried out, a plurality of targets exist in a view field, the targets are required to be marked according to target information, and meanwhile, the mark information which is finished is shared in tracking equipment; (3) Self-adaptive task allocation, namely allocating all targets in the field of view to tracking equipment through a self-adaptive task allocation mechanism; (4) Guiding the tracking equipment to complete tracking of the target in a cooperative working mode of the multiple tracking equipment; (5) After the current task is completed, the tracking equipment participating in the current task performs new task allocation; The step (3) is that the devices communicate with each other, when a target is seen in the field of view of the devices, the position of the target under the geocentric coordinate system is calculated firstly, meanwhile, whether the current target is tracked by the existing devices is identified by comparing the communication between the current target and the adjacent devices, and a plurality of processing modes are allocated for the task of the identification result: 1) When the target is found to have no device tracking When the current target is found to be not tracked by the equipment, executing the step (2) to label the current target, and informing the equipment to track the target; 2) When the target is found to have device tracking When the current device is found to have target tracking, the device needs to judge whether the tracking device of the current target is more suitable than the device, and the tracking of the target is completed, and a scoring function of the fitness of the current target is as follows: Wherein, the Representing the tracking score of the current device for the current target, Representing target information including target brightness and target area information; and (3) if the score of the current equipment is higher than that of the existing tracking equipment, the step (2) is executed to update the label information of the target, when one equipment finds that the label information of the self tracking target is changed, the next target is searched in the field of view, and if the score is lower than that of the current tracking equipment, the next target is searched.
2. The method of claim 1, wherein in step (1), the tracking device aligns the tracking device with the target direction according to the initial guiding information, and waits for the target to enter the field of view of the tracking device.
3. The method of claim 1, wherein in step (2), a labeling mechanism of the object is designed by position information and image information of the object.
4. The method of claim 1, wherein in step (3), tracking devices are allocated to the targets according to their performance and target states.
5. The method of claim 1, wherein in step (4), all the targets are assigned to the tracking devices, and the tracking devices assigned to the same target form a temporary group to track the targets in a coordinated manner.
6. The method of claim 5, wherein in step (5), different tracking devices are allocated to respective tracking targets during tracking of the group targets, after the target allocation is completed, the tracking devices forming the temporary group cooperate to complete tracking of the common task target in a mutual communication manner, and when the current tracking task is completed, the tracking tasks of the rest groups are still performed, and at the moment, the tracking devices need to search new targets again, and add the new targets to the new groups to complete a new round of tracking task.

Description

Group target tracking method by cooperation of multiple tracking devices Technical Field The invention relates to the technical field of collaborative tracking and the technical field of distributed multi-agent collaboration, in particular to a group target tracking method for multi-tracking equipment collaboration. Background With the development of tracking technology, more and more devices are applied to target tracking, and tracking tasks become more and more complex, wherein a task form from single target to multi-target tracking has a specific application scene in various application fields. However, an increase in the number of targets causes two problems: The first traditional single tracking device operation mode is not capable of meeting the requirements; Second, when there are multiple devices in the tracking system, how to integrate the resources of all devices. The cooperative working mode is similar to the distributed working mode of the sensor network, all sensors transmit information in the whole network in an information diffusion mode, and each sensor is combined with the information of all neighbor nodes to locally complete information processing fusion, so that the follow-up work required to be completed is known. Distributed networks have three main advantages: The first concept of removing the central unit in the distributed network, the tasks originally completed by the central unit are dispersed to each local unit, so that the computing resources of the central unit can be reduced; The robustness of the second network can be increased, and the damage of a single node can not greatly affect the network, because other neighbors can take over to complete subsequent work; the expandability of the third network is enhanced, the distributed network operates mainly by mutual communication among nodes, when a new node is added into the network, the communication with a neighbor can know what work needs to be completed, human participation is not needed, and all the distributed work advantages are introduced into the equipment collaborative tracking and exist. However, to introduce a distributed idea into a group target tracking task mainly requires solving two aspects of the problem. First, how to perform task planning, how to distribute all objects to the most suitable tracking device is the problem of task planning. Different target characteristics are different, so that the performances in different types of tracking devices can be different, and therefore, the distribution of proper tracking targets according to the performances of the tracking devices is an important direction of task distribution. Secondly, the problem of equipment coordination is solved, and after the targets are divided, the equipment divided into the same targets need to exchange coordination, so that the smooth completion of the tasks is ensured. In the patent, a multi-device collaborative group target tracking algorithm is designed, so that the problems in the two aspects can be effectively solved, and the working capacity of the system is improved. Disclosure of Invention The invention aims to provide a multi-equipment collaborative group target tracking method which is used for multi-target collaborative tracking, and the multi-tracking equipment resources are organically integrated through task planning, so that the working capacity of a system is improved. The technical scheme adopted by the invention is that the group target tracking method with the cooperation of the multiple tracking devices comprises the following steps: (1) Initial guiding, wherein each device guides the device to an initial position according to the guiding position in an initial stage of starting a task, and waits for a target to enter a field of view of a device detector; (2) When multi-target detection is needed, a plurality of targets are often present in a field of view, the targets are needed to be marked according to target information, and meanwhile, the information that the targets have finished marking is shared in tracking equipment; (3) The self-adaptive task allocation can effectively allocate all targets in the field of view to the tracking equipment through a designed self-adaptive task allocation mechanism; (4) The guiding equipment completes the following of the target in a cooperative working mode of a plurality of devices; (5) When a single task is completed, the devices participating in the current task perform new task assignments. Further, in step (1), the device aligns the device to the target direction according to the initial guiding information, and waits for the target to enter the field of view of the device. In step (2), in order to effectively and organically combine a plurality of tracking devices in the group target tracking task, labeling the target is the basis of subsequent task allocation, and in this patent, we design a labeling mechanism of the target through position information and image informatio