Image completion with generation of rotated patterns and efficient matching

Image completion is a method for completing missing parts caused by the removal of foreground or background elements from an image in a plausible way. In the conventional method, the missing parts of an image are completed by optimizing the objective function, which is defined based on pattern similarity between the missing region and the rest of the image (data region). However, the resultant image could be deteriorated according to variations of patterns in the data region. Although augmenting the pattern variations in the data region leads to good completion result, the processing time to search the plausible pattern in the augmented data increases explosively. In this paper, we propose an extended method considering rotated pattern variations of data region as well as suppressing the computational time by utilizing local orientations of rotated patterns. The effectiveness of the proposed method is validated by comparing with the conventional method.