Method for judging occlusion in target tracking process

Abstract

The invention discloses a method for judging occlusion in a target tracking process. The method comprises the following steps that step 1, reading a tracking target image sequence, extracting a firstframe of image in the image sequence, selecting a to-be-tracked target frame on the first frame of image, extracting an LBP feature vector LBP1 of the target frame wherein the first frame of image isa selected tracking target image in a completely unshielded state; step 2, reading a next frame of image, and performing coarse shielding judgment on the frame of image; step 3, if the step 2 is judged to be shielding, performing accurate shielding judgment t by utilizing a Bhattacharyya distance; step 4, judging whether the image sequence is the last frame of the tracking target image sequence ornot, and if so, ending; and if not, returning to the step 2. According to the invention, a more robust occlusion coarse judgment method is designed, the concept of the Bhattacharyya distance is combined, the interference of target deformation on occlusion judgment is eliminated, and whether the target is occluded or not can be judged more accurately.

Description

technical field [0001] The invention relates to a method for judging occlusion, belongs to the field of target detection and tracking, and is applicable to target occlusion problems in target tracking aspects such as drone tracking and video monitoring. Background technique [0002] Target tracking technology is a technology that has only been developed and put into application in recent years. It is also an important topic and research hotspot at present. Movement information such as speed, direction and trajectory of the moving target can be analyzed and understood in order to complete more advanced tasks, such as missile guidance, military UAV reconnaissance, road traffic monitoring, intelligent monitoring and other technologies are inseparable Target Tracking. [0003] One of the main challenges faced by the target tracking technology based on visible light images comes from target occlusion. In order to solve the occlusion problem, it is necessary to be able to accura...

AI Technical Summary

Problems solved by technology

In order to solve the occlusion problem, it is necessary to be able to accurately judge the occlusion of the target. There have been a lot of researches on the problem of target occlusion judgment at home and abroad. From the initial correlation filtering algorithm, people found that when occlusion occurs, the filter response will appear. The downward trend, that is, the confidence level decreases. Through the change of confidence level, it can be judged that the target is occluded. This occlusion judgment method is simple in principle, but the judgment effect is not good, and it will be affected by many other factors.

In the article "Scale Adaptive Correlation Filter Tracking Based on Occlusion Detection", the occlusion detection method is improved, and the tracking target is divided into four rectangular blocks with the center as the origin, and the peak response (Peak-to-Sidelobe Ratio, PSR) of the four blocks is calculated and analyzed. ) to judge the occlusion of the target. When the PSR of a certain block is less than a certain threshold, it means that there is serious occlusion. This method improves the accuracy of occlusion judgment to a certain extent, but still does not rule out the impact of other changes in the target on the peak response. , and the resulting occlusion misjudgment

[0004] At present, most of the occlusion judgment methods start from the change of the confidence level alone, without considering the impact of the change of the target itself on the confidence level. More research focuses on solving the problem after the occlusion occurs.

Images