Texture Mapping, Texture Compression And Texture Compression

Texturing technique is one of the most popular techniques in computer graphics. It can efficiently simulate or enhance the visual richness of 3D surfaces without increasing their geometric complexity. Texturing technique has been widely applied in many prosperous fields, such as digital entertainment, virtual reality and CAD. In this paper we focus on three key problems in texturing, which are texture mapping, texture synthesis and texture compression. The main contributions are as follows:1. Texture Mapping We propose a novel method for texture mapping based on a probabilistic embedding model which is referred to as Bernoulli embedding model. Our model departs from preserving some relations, such as neighbor relation and distance relation, between pairwise points before and after embedding. We assume that the pairwise-point relations follow Bernoulli distribution over three and two dimensional spaces respectively. By maximizing the similarity between two relational matrices defined over the two spaces before and after embedding, our method efficiently maps the 3D surface points to 2D texture plane with minimum distortion. In order to guarantee valid parameterization, we adopt stretch-based parameterization to optimize the above flattening results. Only a few iterations are enough to converge to valid parameterization with minimal distortion.? We present a new algorithm for constrained texture mapping based on radial basis function (RBF). This algorithm obtains the analytic texture mapping function which passes through the constraints by interpolating the constraints defined by the user with RBF. The energy-minimization property of RBF also ensures that the mapping function smoothly interpolates the constraints to produce minimal deformation. Compared by previous algorithms for constrained texture mapping, our method is fast and can make the mapping results precisely satisfy the constraints.2. Texture Synthesis A novel method for texture synthesis based on user-control is presented. This method is applicable to both arbitrary planar regions and 3D triangular mesh of any topology with convenient control of smooth changes in directions and scales shown in texture synthesis. As for arbitrary planar regions, we decompose them into relative uniform triangular meshes to generate triangles as the unit for texture synthesis. With user-provided vectors which represent directions and scales of texture, the RBF interpolation method gives a smooth vector field which is used to guide the process of texture synthesis. We extract the mask for each triangle to be textured as the search constraints forbest fit. It is natural to extend this algorithm to 3D triangular mesh and use triangles as units for texture synthesis. The texture coordinates for each vertex are outputted as the synthesis result, which saves a lot of memory. Actually, our method gives the unified framework for 2D and 3D texture synthesis. The results show that our method can produce various satisfactory effects of texture synthesis.3. Texture Compression We propose a novel encoding algorithm for texture image ? incremental texture encoding algorithm, in which the compressed textures are decompressed and rendered in real-time by programmable graphics hardware. This method gradually fills in the codebook by dynamically adding in the elements when current codebook can not represent the processed region of the image. In this way, the method can not only produces high compression ratio for texture images with strong self-similarity, but also encode the image sequence in a stream-like manner due to the property of dynamically updating the codebook. Experimental results show that our method is efficient in encoding and rendering still/animated textures. We present an importance-driven texture encoding algorithm based on samples. Our algorithm determines a set of samples from source texture based on combined criteria which include compression ratio, visual attention and parameterization distortion. The sample set i...