A fast texture synthesis using Gene Expression Programming

In computer graphics, vision, and image processing, the texture synthesis occupies an important position. However, most of the existing methods are relatively inefficient. Thus, it is necessary to design a fast texture synthesis algorithm. This article describes an efficient algorithm for texture synthesis. The texture model of this algorithm is derived from the Markov random field. It uses Gene Expression Programming (GEP) to find the best function and the value of each pixel in the synthesized image determined by this function. In this way, this algorithm can avoid scanning all of the pixels, so as to improve the speed of texture synthesis. This algorithm is faster than the previous synthesis algorithm.