Research Of Texture Mapping Based On Multi-directional Image Three-dimensional Model

With the continuous development and progress of science and technology,the use of three-dimensional laser scanning technology for model construction is widely used in cultural relics of protection,ancient architecture and virtual reality extensive of application.With the continuous updating of data acquisition equipments,three-dimensional laser scanning technology is also rapidly improved,and the accuracy requirements of three-dimensional models are becoming higher and higher.At this time,there are more and more researches on texture mapping of three-dimensional point cloud models.How to get more realistic and virtual of high-fidelity scene are also focus on a research of texture mapping.The high accuracy of three-dimensional point cloud models and the realistic of texture information are two important directions in the research of model texture mapping.Texture mapping is the premise of three-dimensional model fidelity and the main task of texture mapping is to deal with the registration of texture images and three-dimensional models.The registration between the texture images and the three-dimensional point cloud models are actually a typical two-dimension and three-dimensional question of matching,and the essence of registration is the texture image and three-dimensional model the extraction and matching of type features.How to map texture through texture images and three-dimensional point cloud models to get more realistic virtual scene has been a problem that we need to solve urgently.In this paper,the methods of feature point extraction and matching about three-dimensional laser point cloud model and texture image are compared and analyzed,and the optimal the feature points extraction of the algorithm method(SIFT)and under the framework of this algorithm of BBF feature point matching are selected,it also prepares for the subsequent of texture mapping.In order to solve the problem of uneven texture and uneven texture distribution caused by overlapping of multiple texture images.The best texture image is selected from the object perspective method,texture mapping is better than multi-image fusion method and the texture shadow elimination algorithm is combined to eliminate the model occlusion,so as to avoiding the holes of the model.In this paper,the experimental process of texture mapping is based on Microsoft VC++ and the framework is based on OpenGL.The experimental results show that the texture mapping experiment can effectively solve the problem of texture blurring after selecting the optimal texture based on the overlap of multiple texture images,it improves the authenticity and integrity of the three-dimensional point cloud model.