Multi-view animal group tracking method and device

Abstract

The invention discloses a multi-view animal group tracking method and device, and the method comprises the steps: obtaining the internal parameters and external parameters of a camera; solving a basicmatrix according to the internal parameters and the external parameters, and calculating the epipolar constraints of different viewpoints; removing a single-viewpoint detection error region accordingto different viewpoint antipodal constraints, and detecting animal skeleton features to estimate an animal motion direction; distinguishing different moving animal identities IDs according to the animal posture characteristics to obtain a tracking result. According to the tracking method, the tracking accuracy and reliability can be effectively improved, the robustness is high, and the tracking method is simple and easy to implement.

Description

technical field [0001] The invention relates to the technical field of visual target tracking, in particular to a multi-viewpoint animal group tracking method and device. Background technique [0002] During the movement interaction process of animal groups, there are serious and frequent mutual occlusion situations. Because of the limited observation angle, the single-viewpoint tracking method will lose the animal area and cannot maintain the continuous movement track of the animals. [0003] In related technologies, the tracking method based on the detection frame does not consider the posture information of the tracking target, resulting in inaccurate predicted motion information. It only searches by expanding the retrieval area. ID (Identity document, ID card identification number), only based on the appearance of animals can not be distinguished, can not guarantee the accuracy and reliability of tracking, which needs to be resolved urgently. Contents of the invention ...

AI Technical Summary

Problems solved by technology

[0002] During the movement interaction process of animal groups, there are serious and frequent mutual occlusion situations. Because of the limited viewing angle, the single-view tracking method will lose the animal area and cannot keep the animal area. continuous motion track

[0003]Related technologies, based on the detection frame tracking method, do not consider the posture information of the tracking target, resulting in inaccurate prediction of motion information, and only search by expanding the retrieval area. Once the apparent In the process of similar animals moving adjacent to each other, when distinguishing the animal ID (Identity document, ID card identification number), it is impossible to distinguish only based on the apparent characteristics of the animals, and the accuracy and reliability of tracking cannot be guaranteed, which needs to be solved urgently

Images