Texture Synthesis From Samples

Texture synthesis has become hot research in Photorealistic Rendering (PR), particularly in the areas of Computer Graphics (CG), Computer Virtual Reality (CVR) and Image Processing (IP) since its inception. Texture Synthesis from Samples (TSFS) is one of the new texture synthesis techniques in recent years. It obtains texture feature data by sampling to establish mathematical models and optimizes these models as tools to synthesize continuous and analogical texture in view.This article makes thorough research in TSFS after summarizing some existing algorithms, some effective solutions are proposed in order to pursuit high quality and high efficiency, specific jobs are as follows:Firstly, analyzed the basic model of two-dimensional texture synthesis, reviewed the development history of TSFS technology and also compared the performance advantages or disadvantages of some classic algorithms.Secondly, An algorithm of fast texture synthesis by block tiling is proposed based on the method of TSFS in order to avoid the defects of seam aliasing by texture mapping technique and debugging complex parameters in procedural texture synthesis. This algorithm uses the correlation technology of position offset to get the mapping addresses of synthesized blocks in the sample image, and integrates the method of searching matching block along a spiral path. Then adopts block boundary matching algorithm to find the best matching block which meets the threshold, and continues this process in a recursive manner until the synthesizing image is fully filled. Experimental results show that this algorithm has achieved better desired effect and used much less synthesis time when dealing with the random texture and the structural texture compared with the previous algorithms.Thirdly, the strategy of searching texture blocks is modified to use some kinds of intelligent algorithms in order to make global search, the block matching algorithm of Wang Tiles is also improved, and the objective evaluation function is combined by SSD, cost of minimum path and the pixel intensity values which greatly improves the performance of the algorithm and the quality of texture synthesis.