Abstract:

The virtual viewpoint synthesis technology based on depth images is an emerging cross-technology in 3D video information processing, virtual reality, and computer graphics. First, the principles of 3D video virtual viewpoint rendering（or synthesis） are introduced. Next, the generation mechanism of the occlusion, artifacts, and holes generated by the virtual viewpoint synthesis is analyzed. Then, the classifications and summarizations of the process of filling holes based on the background reconstruction, image completion, and the restored method based on related images, and the methods to improve the quality of the composite image are listed. Finally, it summarizes and prospects that deep learning and other technologies are the future research directions of virtual view synthesis technology.

Key words: virtual viewpoint synthesis, 3D video images, image artifacts, image holes

摘要：

基于深度图像的虚拟视点合成技术是三维视频信息处理、虚拟现实和计算机图形学领域的新兴交叉技术。介绍了三维视频虚拟视点绘制（或合成）技术的原理；分别就虚拟视点合成过程产生的重叠、伪影、空洞三种问题的产生机理进行分析总结；分类综述了空洞填补中基于背景重建的方法、基于图像补全的方法和基于相关图像的修复方法，并列举了提升合成图像质量的方法。总结并展望了深度学习等技术是虚拟视点合成技术的未来研究方向。

关键词: 虚拟视点合成, 3D视频图像, 图像伪影, 图像空洞