The Research On Texture Reconstruction Based On Multi View Geometry

Every 3D object has rich texture details on its surface. Texture reconstruction is a process of acquiring, processing and mapping the texture details so as to reconstruct the original details on the surface of its corresponding 3D geometry model. It is one of the important fields of photorealistic modeling research. With the rapid development of the optical 3D measurement technology and computer vision, people have a higher requirement for 3D visualization. To model photorealistic 3D objects, firstly, we need to obtain the 3D geometry model of the target object, and then map the corresponding texture images to the model. In order to eliminate the texture mutation caused by factors such as illumination changes on the texture overlap regions, we need to process and blend texture images from different perspectives, and then map it to the 3D geometry model by means of texture blending technology. In this way, texture color transits naturally and smoothly, which contributes to the photorealistic images.This paper firstly introduces the general process of 3D modeling and related technology, with the texture mapping and texture blending basic principle, research status at home and abroad displayed. Then this paper presents the basic theory of multiple view geometry reconstruction, mainly involving camera model, the relationships between different geometric images and triangulation principle, and so on. Then this paper improves and realizes a new texture reconstruction strategy by using the range data and texture image acquired by the optical measurement instrument, combining with range data match technology. This essay adjusts the boundry triangle number using a greed algorithm, to make it as little as possible. Consequently the adjacent triangles have the same texture image. Then the internal triangle and the best texture image are mapped. Next, the smooth transition of the texture color of the frontier triangle is achieved by a kind of composite weights texture fu sion method. Finally, this paper reconstructs the texture of the chinaware. Thus the validity of the proposed algorithm is testified.