Exemplar Image Completion Based on Evolutionary Algorithms
Kangshun Li, Zhen Yang · 2013
As a hot research topic in area of image restoration, image completion has important significance, Criminisi algorithm is one of the most representative completion algorithms. In this paper, according to the inaccurate completion order and the inefficiency in searching matching patches used in Criminisi algorithm, a exemplar image completion based on Evolutionary Algorithms is proposed. The information of gradient direction and isophote direction decides the order of image completion, and choose the appropriate size of template window according to differences in the frequency. Then the algorithm can restore image quickly and accurately by best matching patches which obtained through Evolutionary Algorithms which search the optimal patch in the area around the damaged region. Experimental results show that the proposed algorithm has better general performance in image completion.