Robust principal component analysis via weighted nuclear norm with modified second-order total variation regularization

The traditional robust principal component analysis (RPCA) model aims to decompose the original matrix into low-rank and sparse components uses nuclear norm describe prior information of natural image. In addition low-rankness, it has been found in many recent studies that local smoothness is also crucial low-level vision. this paper, we propose a new RPCA based on weight modified second-order total variation regularization (WMSTV-RPCA for short), which exploits both global low-rankness matrix. Extensive experimental results show, qualitatively quantitatively, proposed WMSTV-RPCA can more effectively remove noise, dynamic scenes compared with competing methods.