Texture Synthesis Based on Directional Empirical Mode Decomposition

Texture is an essential aspect in virtual reality,which makes texture synthesis an important research.This paper proposes a novel texture synthesis algorithm.In this algorithm,the signature direction of a sample texture is obtained by using DEMD (direction empirical mode decomposition) method.Then the calculated direction is processed into a set of decomposed images by using direction EMD approach.The feature values extracted from all the levels of the decomposed images are used to search matching patterns.The synthesized image is formed by using a correlation searching.Thus a synthesis associated with the direction is made after composing all levels of decomposed images.In the end the final image can be synthesized by rotating the image along the calculated direction.Our experiment results show that this algorithm is simple, fast and delivers high quality synthesis.