Study On Natural Spectacle' Texture Synthesis Based On Neural Network

Texture synthesis is one of the important tasks of Digital Image Processing and Computer Graphics, which have been developed quickly in recent years. It can be extensively used in the fields of communication, security, etc, so it is of great value. To solve the present problems in texture-image synthesis, this project makes use of the more mature Artificial Neural Network BP algorithm in artificial intelligence, so that the texture synthesis of natural spectacle is accomplished smoothly. The task mostly includes: study and analyze the eigenfunctions of various textures and select 16 eigenfunctions ; set up an effectual Artificial Neural Network BP model by way of the numerical values of the above 16 eigenfunctions as its inputs; finally, form the program of natural texture synthesis. Giving a natural texture image, the program will extract the corresponding eigenfunction automatically and set up its Artificial Neural Network structure, thus finally realize the texture-image synthesis. There is an experiment showing that this method breaks up the limitations in the traditional methods, that is the texture-image is synthesized according to its structural model. In addition, it can apply various texture images.