Three-dimensional Texture Classification Based On Wavelet Packet

Texture classification is an important branch of pattern recognition. The primary purpose is that distinguish different feature and presentation of textures. It has a wide background of computer vision, computer graphics and engineer technology. The collection, analyzing and classification of 3D texture become one of the keen topic in this field. Now the discussion of 3D texture classification is an advanced orientation with great opportunity.The mission of texture classification is determining the same kind with similar features, which can relieve human beings from tremendous hard work while observing so many pictures, furthermore, the intelligent method can reach a high rate of right recognition. Up to date, technologies of 2D texture classification lead a fast development. But unfortunately, the input and output are 2D at the same time, so they may lose their power to transfer the usage to 3D texture classification. Moreover, the algorithms can not create effective images at different lighting angles.Our work begins with 2D texture classification and finishes all kinds of 2D classification algorithm as a system firstly. after that, we purpose a simplified Bayes classifier based on wavelet packet in the algorithm that used in the oil project, the optimal parameter are listed in the result. Secondly, according the trait of kernel function, conjunction with texture classification, we purpose a classification algorithm based on kernel function. So we extend the idea to 3D texture classification research. At last, a plenty of experiments tell us the algorithm is effective and meaningful.