Tracking method based on sparse subspace

Abstract

The invention discloses a tracking method based on a sparse subspace, comprising the steps of firstly learning a plurality of initial frame images by using a random projection matrix and a robust principal component analysis (RPCA) method, and obtaining a low-rank matrix of the images; and extracting the sparse subspace where a tracked target is located from the low-rank matrix. The sparse subspace obtained according to the method has the characteristics of low complexity and high robustness. Compared with the traditional particle filter method based on the target color, target texture or target template, the algorithm based on the target features mentioned in the text has the characteristics of being less in the number of required particles, high in timeliness and strong in stability.

Description

technical field [0001] The invention relates to the technical field of pattern recognition, in particular to a tracking method based on a sparse subspace. Background technique [0002] Target tracking mainly solves issues such as mobile signals, object positioning and background estimation that change over time, and is widely used in video surveillance, human-computer interaction and other fields. Although more successful tracking algorithms have been developed and widely used in recent years, there are still many challenging problems to be solved in this field, such as: the target is less sensitive to optical flow changes; The similarity weakens the tracking effect; when the shape of the target changes, and when the target encounters severe occlusion, the tracking algorithm is less robust; it can only adapt to a lower degree of occlusion. Therefore, extracting the stable features of the target relative to the above various changes during the movement process plays a vital ...

AI Technical Summary

Problems solved by technology

The more common features are pixel intensity, color, texture, etc. in the target area. Although these target features can be used as important tracking elements to a certain extent, due to lack of robustness, the tracking algorithm based on these features cannot adapt to the target area. Various serious changes encountered in sports

[0003] In recent years, a considerable part of the research on target tracking and detection uses SIFT features as the main extraction factors, but SIFT has the following disadvantages: due to the need to build image pyramids, the amount of calculation is usually relatively large; SIFT features are essentially A local feature, when the tracked target encounters severe occlusion, the program will lose the target due to lack of robust features; when the target appearance changes, the original SIFT will be lost

However, this method has two main deficiencies: the method of obtaining the principal component analysis through the singular value decomposition of the matrix has a large amount of calculation; in the case of occlusion, the method cannot track the target stably

Images