Research On Texture Synthesis From Sample

The texture synthesis technique based on the sample image is to synthesize a large area texture image from the texture features of the given sample image and the existing texture structure features and ensure the similarity or continuity of the texture structure.This technique searching for matching points in sample textures and copying them to output images to generate continuous textures similar to the samples are widely used in many areas,such as realistic or non-realistic rendering,image restoration and computer-aided design.The paper mainly improves the block collage texture synthesis algorithm to improve the quality and speed of texture synthesis results.In addition,the improved texture synthesis algorithm is applied in the process of texture transfer,completed the conversion of artistic style between different images.The research work done in this paper is as follows:An improved texture synthesis algorithm based on multi-order error surface is proposed to deal with the boundary discontinuities caused by the collage errors in image quilting texture synthesis algorithms.By changing the order of the error surface formula to obtain a more accurate optimal cutting path,the structural information during the splicing is maintained better.The experimental demonstration and data analysis show that by increasing the order of the error surface formula,we can get a more accurate cutting path to obtain better synthesis result.In order to further improve the quality of the synthesis results of the block collage algorithm,this paper optimizes the seams after the collage.After finding the best cutting path to complete the collage of texture blocks,poisson blending is introduced to smooth the discontinuity of joints in texture synthesis.The experimental results show that the discontinuous region of the original algorithm is improved by poisson blending optimization.Besides,the improved algorithm is extended to the implementation of texture transfer and acquired good results.In order to solve the problem that the global search method of matching block is time-consuming,an optimal search algorithm based on genetic evolutionary search strategy is proposed.By using genetic search strategy to search the best matching block,the search speed is accelerated and the efficiency of the algorithm is improved.The experimental results show that the improved algorithm has greatly reduced the search time and achieved better results.