The Application Of Texture Synthesis In Video Synthesis

Sample-based texture synthesis is a new texture synthesis technique developed from the texture mapping and procedural texture synthesis. It is efficient to solve the problem which in traditional techniques, such as noticeable seams between texture patches and the stretch and distortion of pattern when tiling a texture on surfaces. This article mainly studies the application of texture synthesis technique in vector field visualization and video synthesis.This paper starts by introducing several basic concepts on texture synthesis. After that, with a detailed description on typical algorithms, current research status of texture synthesis is discussed and key issues are also pointed out. In the applications of vector field visualization, current texture synthesis algorithms have the problem of that the angle and length of inside point vectors are different from each other greatly and they can't show the trend of vector field properly. To solve this problem, my paper presents an improved hybrid algorithm which use both block-based and point-based synthesis as adaptive segmentation. Under the frame of constrained texture synthesis, the article defines the correlate constraining of points in two adjacent frames guided by vector field, uses texture synthesis to generate continues frames and implements the vector field visualization by video. In comparing the visualization by generating images, the video way shows the field transformation more directly. In the application of video generation, current synthesis based algorithms are introduced. This paper also gives an approach on single image video generation which extracts the flow part of input image by segmentation algorithms and forms the movement track by interaction. In order to accommodate different vector field, the sample image is rotated and zoomed. This paper defines the correlate constrain on multi-frame, and synthesize a group of consecutive frame by this constrain, and achieve video synthesis only use single image.Our research can be used in computer animation, large scale scene rendering, Simulation of the scenes and so on.