Super Resolution Of 3D Surface Texture Based On General Training Set

Texture is a measure of the variation of the intensity of a surface, quantifying properties such as smoothness, coarseness and regularity. It's often used as a region descriptor in image analysis and computer vision, and is an important method of figuring real world. The representation, synthesis and editing of 2D surface textures, play an important role in computer vision, virtual reality, medicine and spaceflight. In fact, the resolution of an image lies on how many details it contains. In order to enhance the resolution of an image, the prevailing method is predicting the lost details of an image by the technique of super resolution.The super resolution of 2D surface texture has gained a lot, but the real-world surface textures normally comprise rough surface geometry and various reflectance properties. So the super resolution based on 2D texture cannot meet our need of rebuilding the real world, and a new method which can deal with the 3D surface texture is demanded. For example, if you map a 2D texture on a rabbit when you modeling the real world, you can't achieve the changes of feathers when the light angle and view angle change in real-time. Therefore, it has an important realism of doing research on super resolution about 3D surface textures.Based on the previous simple super resolution technique of 3D surface textures, this paper develops its super resolution of 3D surface textures based on eigen-base images. Next, this paper puts up a new super resolution about 3D surface textures based on general training set by combining the Textons with SVM theories, by which the super resolution of 3D surface textures is made easier and more practical. At last a set of experiments and results are made on the evidences of the efficiency of the algorithm, which is proposed on super-resolution of 3D surface textures.