Key Technologies of Seam Fusion for Multi-view Image Texture Mapping Based on 3D Point Cloud Data

Abstract

With the rapid development of computer technology and measurement technology, three-dimensional point cloud data, as an important form of data in computer graphics, is used by light reactions in reverse engineering, surveying, robotics, virtual reality, stereo 3D imaging, Indoor scene reconstruction and many other fields. This paper aims to study the key technology of 3D point cloud data multi-view image texture mapping seam fusion, and propose a joint coding and compression scheme of multi-view image texture to replace the previous independent coding scheme of applying MVC standard compression to multi-view image texture. Experimental studies have shown that multi-view texture depth joint coding has different degrees of performance improvement compared with the other two current 3D MVD data coding schemes. Especially for Ballet and Dancer sequences with better depth video quality, the performance of JMVDC is very obvious. Compared with the KS_ IBP structure, the gain can reach as high as 1.34dB at the same bit rate.