Texture Synthesis Based On Model Segmentation And Parametrization

Model texturing technique is one of the most popular techniques of realistic graph rendering. It can efficiently simulate the details of 3D surfaces without increasing the complexity of the geometric model, so it has been widely applied in many fields such as 3D game, computer animation, virtual reality, computer aided design (CAD) and so forth. In this thesis, we deeply research on some model texturing techniques, such as parameterization, 3D model segmentation and texture synthesis from samples.Firstly, we research on a question of surface parameterization. To resolve the texture mapping for texture distortion problem, we propose a novel texture synthesis method. Based on the relation of the neighboring triangulated surfaces of model, we use an extent preferential search strategy to appoint texture coordinate of each triangulated surface. In the processes of searching and mapping have restricted surfaces, First, we search a minimal texture error of restricted surfaces, then, joint the textures into a sequential macrocosm by using graph cut technique.Secondly, a normal evaluation method of segment model for texture synthesis is proposed. Based on the vector and mean curvature of the neighboring triangulated surfaces of model, we use extent preferential search strategy to segment the model. We use orthographic projection mapping to parameterize and evaluation for each model block. When the model block has large distortion, we can fractionize the block again until distortion value achieves the threshold value. Finally, we use degree of similar accumulation to unite some of model block. The method is combined theory of parameterization and applied in model texturing rendering.The research of the model texturing techniques indicates the two methods proposed by the thesis not only keep the characteristic of texture, but also have better treatment for the seam of texture. Therefore, our methods are efficient for model texturing rendering.