Structure and Texture Image Inpainting Based on Region Segmentation

In this paper, we propose a new algorithm for structure and texture filling-in of complex images with missing information. Our algorithm relies on edge-based region segmentation. The segmented regions are used both to reconstruct a structure component and to guide the restoration of a texture component. The contributions of this paper are two-fold. Firstly, we propose an efficient method to prevent the edge-blur in filling-in complex image. Secondly, the texture can be quickly and nicely fixed in our method. Examples on real images show the advantages of our algorithm.