Non-Local Sparse and Low-Rank Regularization for Structure-Preserving Image Smoothing

Lei Zhu1     Chi-Wing Fu1     Yueming Jin1    Mingqiang Wei2    Jing Qin3    Pheng-Ann Heng1

1The Chinese Univeristy of Hong Kong       2Hefei University of Technology       3The Hong Kong Polytechnic University


More results produced by our method. Top row: input images with different texture patterns. Bottom row: our results. From left to right: the first input image courtesy of flickr user Peter M¨ arz, while the other two are from the public data set(Link).


This paper presents a new image smoothing method that better preserves prominent structures. Our method is inspired by the recent non-local image processing techniques on the patch grouping and filtering. Overall, it has three major contributions over previous works. First, we employ the diffusion map as the guidance image to improve the accuracy of patch similarity estimation using the region covariance descriptor. Second, we model structure-preserving image smoothing as a low-rank matrix recovery problem, aiming at effectively filtering the texture information in similar patches. Lastly, we devise an objective function, namely the weighted robust principle component analysis (WRPCA), by regularizing the low rank with the weighted nuclear norm and sparsity pursuit with $L_{1}$ norm, and solve this non-convex WRPCA optimization problem by adopting the alternative direction method of multipliers (ADMM) technique. We experiment our method with a wide variety of images and compare it against several state-of-the-art methods. The results show that our method achieves better structure preservation and texture suppression as compared to other methods. We also show the applicability of our method on several image processing tasks such as edge detection, texture enhancement and seam carving.


Supplementary Material

[More results]



We thank reviewers for the valuable comments. This work was supported by the Hong Kong Research Grants Council General Research Fund (Proj. No. 412513), the National Natural Science Foundation of China (Proj. No. 61233012, 61502137), and the CUHK strategic recruitment fund and direct grant (4055061).



Snapshot for paper "Non-Local Sparse and Low-Rank Regularization for Structure-Preserving Image Smoothing"
Lei Zhu, Chi-Wing Fu, Yueming Jin, Mingqiang Wei, Jing Qin, and Pheng-Ann Heng. Computer Graphics Forum, 35(7)(Proc. Pacific Graphics), 2016

  [Paper (pdf, 20.9MB)]

 [Matlab Code (zip, 6.33M)]



inproceedings {zhu2016non,

       title = {Non-Local Sparse and Low-Rank Regularization for Structure-Preserving Image Smoothing,

       author = {Lei Zhu, Chi-Wing Fu, Yueming Jin, Mingqiang Wei, Jing Qin, and Pheng-Ann Heng},

       journal = {Computer Graphics Forum (Pacific Graphics 2016)},

       volume = {35},

       number = {7},

       pages = {217 - 226},

       year = {2016},

       organization={Wiley Online Library},






Last update: March 29, 2017