Video Moving Objects Detection by Minimizing the Partial Sum of Singular Values of the Matrix

Abstract

As an important part of computer image processing technology, moving object detection has been applied more and more widely in real life, whether it is security monitoring in public places, artificial intelligence behavior recognition, or even military fields. Robust Principal Component Analysis (RPCA), also known as Low Rank and Sparse Decomposition (LRSD), is widely used in moving target detection. In a low-rank matrix decomposition problem, we not only know that the underlying structure is low-rank, but also know the condition of the exact rank of the matrix data. In the rank constraint problem, the prior information of the target rank can be fully utilized in the expression of the objective function, which can solve the problem better. The innovation of this paper is to use this information to improve the robustness of object detection in various complex scenarios. Finally, the effectiveness of the algorithm is verified by experimental results.