Tracking research method based on dynamic/static target detection and real-time compressive sensing

Abstract

The invention discloses a tracking research method based on dynamic / static target detection and real-time compressive sensing. Self-adaptive mixed Gaussian foreground extraction is utilized to extractforeground information from current backgrounds, then expanding corrosion of morphological processing is utilized to enable regions, which are originally in noises, to be removed, and finally, targets are separated from the backgrounds by a block analysis tool, and the number, positions, shapes, directions and sizes of the targets are calculated; static detection is taken for targets which are stationary for long time, local-binary-model texture features are extracted from images, and are sent into a cascaded classifier to carry out training and learning thereon to obtain positions of the static targets; and finally, current targets are ultimately tracked in real time through real-time compressive-sensing tracking based on correlation filtering, solution optimality and lower time complexity of correlation filtering are utilized, and motion trajectories of fishes are ultimately determined. According to the method, negative impacts of interference of complex environments on target detection tracking results are reduced, an effect of real-time detection tracking is achieved, and robustness of an algorithm is enhanced.

Description

technical field [0001] The invention relates to a method based on dynamic and static target detection and real-time compressed sensing target tracking, and belongs to the technical fields of intelligent information processing and target detection and tracking. Background technique [0002] Object detection and tracking is an integral part of most vision systems. In specific scene applications (such as video surveillance and other fields), automatic, fast, and highly robust object tracking has attracted attention. It has broad application prospects in video surveillance, traffic detection, intelligent robots, and submarine target detection and tracking. [0003] Object detection and tracking is an extremely important part of the field of computer vision. The purpose of moving object detection is to segment the foreground object of interest from the sequence of images, and suppress the background noise and foreground noise as much as possible, so as to accurately obtain the s...

AI Technical Summary

Problems solved by technology

The occlusion and deformation between objects, the complexity of the background, the change of light and dark, poor real-time performance and robustness are all problems that need to be solved in the tracking process.

Classical tracking methods such as Meanshift and particle filter rely on the richness of the target information contained in the video. In the actual video image sequence, the information that the target can provide is quite limited, resulting in the inability to track the target stably. Changing occlusion, these classic algorithms are powerless

[0005] That is, the main problems in the existing technology: (1) The target to be detected has changes such as changes in illumination and deformation, and the movement state of the target to be detected is complex, with movement and stillness, resulting in serious missed detection of the target

(2) The real-time and robustness of the tracking process in the video scene to be tracked is poor, the space-time position information of the target is lacking, and the target features are not obvious; (2) When the scene has occluders and the target to be tracked is deformed, especially There will be a situation where the entire target is occluded and the target to be tracked undergoes a huge deformation, which will lead to the loss of the tracked target

Images