On Schatten-q quasi-norm induced matrix decomposition model for salient object detection

作者：

Highlights：

• A weighted group sparsity induced norm developed in this paper is shown to be able to make the foreground being consistent within the same image patches by effectively absorbing the image geometrical structure as well as the feature similarity.

• The Schatten quasi-norm is successfully used to capture the lower rank of background via factorization technique, and an alternative non-convex formulation for nuclear norm is presented.

• Extensive experiments on the five widely used datasets show that the proposed approach outperforms those by using state-of-the-art models in current literature.

摘要

•A new novel regularization model for salient object detection is proposed, which integrates a weighted group sparsity with the convex Schatten-1 or the non-convex Schatten-2/3 and Schatten-1/2 norm, respectively.•A weighted group sparsity induced norm developed in this paper is shown to be able to make the foreground being consistent within the same image patches by effectively absorbing the image geometrical structure as well as the feature similarity.•The Schatten quasi-norm is successfully used to capture the lower rank of background via factorization technique, and an alternative non-convex formulation for nuclear norm is presented.•Extensive experiments on the five widely used datasets show that the proposed approach outperforms those by using state-of-the-art models in current literature.

论文关键词：Low rank approximation,Schatten-p norm,Matrix decomposition,ADMM,Salient object detection

论文评审过程：Received 17 November 2018, Revised 8 June 2019, Accepted 15 July 2019, Available online 29 July 2019, Version of Record 31 July 2019.

论文官网地址：https://doi.org/10.1016/j.patcog.2019.106975