Research On Sample-based Texture Synthesis With Constraints

After years of study, texture synthesis has been one of the hotspots in the area of Computer Graphics, Image Processing and Computer Vision. This article introduces some improvements on the multi-sample based texture synthesis based on the existing algorithms. It also introduces an improved algorithm on the synthesis of irregular texture.Firstly, this article gives a brief introduction on the texture synthesis technique and introduces some basic concepts of texture synthesis. Then it gives a general view of recent progresses on sample-based texture synthesis technique. Among these progresses, multi-sample based texture synthesis is one of the directions of the future research, and it's also the main concern of this article. Because of the importance of space distribution of texture in human perception of the images, an image segmentation algorithm is introduced into the texture synthesis for texture area selection, and the constraint of the original image is added to keep the coherence of the target image. This method can be used in texture design and image editing. This article also introduces a texture synthesis method for the irregular images which contain some structural information. To overcome the weakness of existing algorithms on those structural but irregular textures, a new texture synthesis algorithm with an adaptive texture unit selection is proposed. By picking up the texture units of the image and using these units as the elements of the synthesis process, this method could gain some improvements in results.Our research can be used in computer animation, large scale scene rendering, pattern design, special effects in movie and video, virtual reality and so on.