Three-Dimensional Surface Texture Synthesis Based On Kernel Function

Texture synthesis has broad application areas, such as the design of cloth, soft furniture packing, various table-board textures, inner decorations of cars, post-processing in film and electronic game industries, as they all need different textures to be synthesized. Texture synthesis is also an active research field in computer vision, image processing and computer graphics.The aim of texture synthesis is to produce an image which is larger than the original one, while it has the same characteristics of the sample texture. Current techniques of 2D texture synthesis are nearly mature. However, as both of the input and output images are 2D intensity texture images, it cannot provide the information required for rendering under other than the original illumination and viewpoint conditions if the subjects are 3D surface textures.Three-dimensional texture synthesis deals with 3D texture synthesis and editing under different illumination and viewpoint conditions. It involves in data representation,extraction, analysis, synthesis and texture rendering. As a state-to-art area in image processing, 3D surface texture synthesis has strong research potential.In this paper, the 2D synthesis methods are first introduced and analyzed. Then the texture synthesis algorithm based on kernel function is proposed by combining texture synthesis and kernel function. It is further extended to 3D texture synthesis. Our experiments show the effectiveness of the proposed 3D texture synthesis algorithms based on kernel function.