| Texture Synthesis from Sample (TSFS) is a new way of Texture Synthesis in nearly decade. It takes small texture image as input to synthesis the arbitrary size texture image, the visual appearance of which is similar and continuous to that of the original image. TSFS eliminates some defects of Texture Mapping technique. Moreover, the boring processes of parameters adjustment of Procedural Texture Synthesis (PTS) can be avoided. Then, it becomes more and more attractive, and to be a research hotspot in Computer Graphics, Computer Vision and Image Processing. In this paper, based on the study of Texture Synthesis from Sample, Dissertation presented improvements of the existing algorithms, the main work include:1. Briefly introduce background and significance, current situation, problems and applications of the Texture Synthesis from Sample, then introduce the basic concepts of texture synthesis based on the MRF model, and study some of the classic point-based and patch-based algorithm synthesis, analyze the thought of the algorithm and the advantages and disadvantages of the algorithm and right type algorithm to be hand.2. A new algorithm is presented. It considered the similarity of the color and gradient. The colors and gradients as a measure of the similarity of between the two matching patch, the extent of the similarity is improved. Experiments show that synthesizing results are quite satisfactory.3. In this paper, a new texture synthesis algorithm is presented. It adopts the color and the gradient as the similarity measurement between the two matching patch, which improves their similarity extent. Moreover, a spiral-path searching method is used to search the best texture matching patches and the synthesis time can be decreased dramatically. The experiment results show that not only the synthesis speed is rapider in great degree, but also the synthesis quality is improved.Our research can be used in computer animation, virtual reality special effects in movies and games and so on. |
PDF Full Text Request