Image texture synthesis method based on the trend function constraint

作者：Yuanying Gan, Chuntong Liu, Zhenxin He, Hongcai Li, Zhongye Liu, Xinling Yao · 发表于：2021 IEEE 3rd International Conference on Frontiers Technology of Information and Computer (ICFTIC) · 年份：2021 · DOI：10.1109/icftic54370.2021.9647387 · 被引用次数：2 · 研究领域：Image Processing and 3D Reconstruction、Computer Graphics and Visualization Techniques、Generative Adversarial Networks and Image Synthesis

Texture synthesis technology can not only be used for texture transmission and repair damaged photos, but also has a broad application prospect in large-scale virtual scene rendering. In order to solve the problems of seam aliasing in texture mapping and considering the self-similarity of texture, this paper defines the trend function based on the biased random walk algorithm, divides the motion unit with different seed points and constructs the adjacency matrix, and calculates by matching labels Random walk stepping probability to realize texture synthesis. Compared with the background image segmentation results, it can be seen that the texture image generated by the algorithm in this paper presents similarity and continuity visually, and shows strong randomness at the same time.