The Research On Image Restoration Arithmetics Based On Texture Synthesis

Image restoration is one of the important research hotspots in computer graphics, computer vision and image processing. It has great potential in many applications, especially in restoration of art, fast data transferring on the Internet, computer animation, special effects, virtual reality and so on. The paper begins with an introduction of the image restoration system. In this paper, the key techniques on image restoration are discussed and analyzed, include inpainting, texture synthesis and image segmentation. Some important algorithms of image restoration are summarized and the trends of them are forecasted. Based on these theory, a robust image restoration system based on texture synthesis is devised and realized.The work includes important aspects as follows: The representative inpainting algorithms are analyzed and summed up, which include local inpainting and statistics-based inpainting. The limitation of inpainting is pointed out. The characteristics and techniques of texture synthesis are discussed and summed up. The ideas and internal relation in logic of important texture synthesis algorithms are summarized, among which patch-based texture synthesis arithmetic is implemented and improved. In addition, the trend that texture synthesis method should be turned into a core of image restoration is pointed out. The typical algorithms of image restoration are analyzed, the characteristics andcapability of them are compared and discussed. A system on image restoration is devised and realized, which is based on texture synthesis and region growing method.