A moving target tracking method based on a matrix low-rank representation theory

Abstract

The invention brings forward a moving target tracking method based on a matrix low-rank representation theory in the framework of a particle filter theory. A moving target observation matrix composed of image sequences is decomposed into a low rank matrix representing a moving target and a sparse matrix representing shields; the low order matrix is used as a moving target model, and the sparse matrix is used as a target shielding model; a target observation probability optimization equation is established by using the moving target model and the target shielding model; the moving target model and the target shielding model are solved separately by using an optimization algorithm; and the moving target model and the target shielding model are substituted into the target observation probability optimization equation to calculate a moving target observation probability required by target tracking so as to achieve the tracking of a moving target. According to the invention, a problem of tracking drift and even target losing after the moving target is shielded is solved, thereby improving the anti-shielding ability of the moving target tracking. The moving target tracking method of the invention has high robustness.

Description

technical field [0001] The invention belongs to the technical field of computer vision processing, and in particular relates to a moving target tracking method based on matrix low-rank representation theory. Background technique [0002] When tracking a moving target, for a given video sequence, the moving target tracking method is to detect, extract, identify and track the moving target to obtain more accurate parameter information such as the position, speed, and acceleration of the target, and pass Further processing can realize the understanding of the behavior of the sports target, and then complete higher-level tasks. It can also be said that the matching of image features such as position, speed, shape, texture, and color is realized between consecutive image frames. Therefore, the key to the moving target tracking method is to establish the corresponding relationship between the candidate target and the reference target in each frame image. However, due to factors ...

AI Technical Summary

Problems solved by technology

However, due to factors such as illumination changes, moving target shape changes, background changes, and moving targets being partially or completely occluded, for a given image sequence, the feature information on the moving target may disappear for a while and then reappear or even completely disappear

[0003] For existing moving target tracking methods, such as IVT method and L1T method, usually only consider the relationship between two adjacent frames of images, while ignoring the continuity between the entire image sequence in the tracking process, when the moving target is partially occluded Or when it is completely occluded or affected by factors such as drastic changes in illumination, the tracking results are prone to drift or even the moving target is lost

Images