Research On Texture Synthesis Algorithm For Multiple Sample

Sample-based texture synthesis (Texture Synthesis from Samples) technology isdeveloped rapidly in recent years, a new texture synthesis technique, which is basedon a small piece of a given sample texture, according to the geometry of the surface,split generates visually similar, continuous throughout the surface texture,performance geometry model can not reflect the details, to compensate for thedeficiencies in geometric drawing, image based rendering and occupies an importantposition in. Sample-based texture synthesis technology not only overcomes thedisadvantages of traditional method of texture mapping, while avoiding theprocedural texture synthesis process of frequent adjustment of parameters, ascomputer graphics, computer vision and image processing fields.Based on texture synthesis methods to study, mainly on sample-based texturesynthesis methods of. The specific research contents are as follows:first briefly introduced the concept of texture synthesis texture, as well as theresearch background and research significance. There are some texture synthesisbased knowledge. And the existing two-dimensional, three-dimensional and surfacetexture synthesis method to carry on the analysis, compare the advantages anddisadvantages.in the study and summarize the existing texture synthesis algorithm based on,the design and implementation of an improved single texture synthesis algorithm,this algorithm based on the texture of neighborhood correlation theory, combinedwith the spiral synthetic sequence and matching block location distributioncorrelation, look for the best matching texture blocks in order to reduce thecomputational complexity, can accelerate the speed of texture synthesis purposes,and through experiments prove the validity of the improved algorithm. Study oftexture synthesis algorithm of three major parameters (texture block size, textureblock number and boundary region size) on texture synthesis quality and speedeffect. On the analysis of the basis of the experimental results, obtained withdifferent texture synthesis demand parameter selection scheme.(3) the diversity of texture synthesis methods research, try based onmulti-sample texture synthesis, presented in a user constraints on input sample texture mixing method, mixing in texture in the process of taking user constraintsand a plurality of input samples of these two factors, using the sample spacedistribution control and sample distribution proportion control two synthesismethods, so that the texture mixed results not only have many samples the structurecharacteristics, at the same time the user set the intuitive constraints do better with.