Image Completion Using Dynamic-Scale Patch Matching and Layer-Wise Chain Optimization
Xu Gan · Dianzi xuebao · 2015
In order to reduce visually inconsistent results caused by sudden change of pixel values betw een patches,a novel image completion method based on dynamic-scale patch matching and layer-w ise optimization w as proposed. During patch searching,the number of candidate patches for the current layer w as calculated through the analysis of prior know ledge and structure features; meanw hile,a multi-scale patch searching model w as given to obtain the best candidate patches. Those patches constituted the feasible solution space for image completion. With the intrinsic characteristics and relevance of adjacent patches taken into consideration,image completion w as abstracted as a chain optimization problem. The layer-w ise chain optimization model w as established and solved through dynamic programming. The optimal patches for the current layer w ere obtained and the image w as repaired from the outside to the inside layer by layer. Experimental results demonstrate both the effectiveness and efficiency of the proposed algorithm for various natural images.