¹The Graduate School of Systems Science and Technology, Akita Prefectural University, Akita, Japan.
²Dept. of Electronics and Information Systems, Akita Prefectural University, Akita, Japan.

---

Abstract—This paper proposes a novel image inpainting algorithm based on sparse representation with histogram dictionary. In the propose method, it is a patch-based inpainting method, a patch as the basic unit of inpainting algorithm. Instead of using all the known image patches to form a dictionary, we propose to construct a histogram dictionary by using the histogram. If the histogram between an image patch which cuts from the source region of the input image and the target patch are similar, the image patch as the similar patch is selected to generate a similar dictionary for inpainting algorithm. Two kinds of similarity methods are proposed, so two kinds of histogram dictionaries are corresponding. These histogram dictionaries can ensure a large related to the target patch. Experimental results show that the proposed inpainting method using the histogram dictionary can efficiently inpainting of missing areas.

Index Terms—Image inpainting, sparse representation, histogram dictionary, similarity.

[PDF]

Cite: Qiaoqiao Li, Guoyue Chen, Xingguo Zhang, Kazuki Saruta, Yuki Terata, "Image Inpainting Based on Sparse Representation with Histogram Dictionary," Journal of Computers vol. 13, no. 10, pp. 1145-1155, 2018.