Int J Performability Eng ›› 2018, Vol. 14 ›› Issue (11): 2777-2788.doi: 10.23940/ijpe.18.11.p24.27772788

Previous Articles     Next Articles

A Novel Image Inpainting Method for Object Removal based on Structure Sparsity

Lei Zhang* and Minhui Chang   

  1. School of Mathematics and Information Technology, Yuncheng University, Yuncheng, 044000, China
  • Submitted on ;
  • Contact: * E-mail address:
  • About author:Lei Zhang received his B.S. degree in computer science and technology from Shanxi Normal University, Linfen, China in 2003, his M.S. degree in computer application technology from Shanxi University, Taiyuan, China in 2008, and his Ph.D. in computer application technology from Northwest University, Xi'an, China in 2016. His research interests include image processing, pattern recognition, and machine learning. E-mail: Minhui Chang received her B.S. degree and M.S. degree in basic mathematics from Shaanxi Normal University, Xi'an, China in 2003 and 2009, respectively. Her research interests include wavelet analysis and machine learning. E-mail:

Abstract: In the traditional image inpainting method for object removal, for each target patch, the entire source region must be traversed to search for the exemplar patch, which may make the restoration process time-consuming and affect the restoration efficiency. Even worse, the target patch may be replaced by an inappropriate exemplar patch during the process, which will introduce some unexpected objects in the restored image and make the result unable to meet the requirements of visual consistency. In view of these problems, we propose a novel image inpainting method for object removal based on structure sparsity. First, we calculate the structure sparsity of the target patch, and then identify the local characteristics of the region where the target patch is located. Then, we set different search regions for the target patches according to different regional characteristics. Finally, we find the exemplar patch in the search region and restore the target patch. Experiments on a number of natural images show that the proposed method can reduce the restoration time and improve the restoration efficiency. Additionally, it can prevent the mismatch to some extent and improve the restoration effect.

Key words: Image Inpainting, object removal, structure sparsity, search region