面向光场显示的虚拟视点生成技术进展  

Research progress of virtual view rendering technology for light field display

作　　者：赵健 戴子尧[1] 丁义权 夏军 ZHAO Jian;DAI Zi-yao;DING Yi-quan;XIA Jun(School of Computer Engineering,Nanjing Institute of Technology,Nanjing 211167,China;School of Electronic Science and Engineering,Southeast University,Nanjing 210096,China)

机构地区：[1]南京工程学院计算机工程学院,江苏南京211167 [2]东南大学信息显示与可视化国际联合实验室,电子科学与工程学院,江苏南京210096

出　　处：《液晶与显示》2023年第10期1361-1371,共11页Chinese Journal of Liquid Crystals and Displays

基　　金：国家自然科学基金(No.62075040);江苏省高等学校基础科学(自然科学)研究项目(No.21KJB520038)。

摘　　要：随着元宇宙产业的发展,光场显示技术因其可以将数字世界与物理世界完美融合,已经成为信息显示技术领域的研究热点,并呈现出视点数量更多、视点密度更大、渲染速度更快的发展趋势。然而,现有的光场显示技术面临着分辨率低、深度受限、视疲劳等多种挑战。本文从人眼立体视觉感知原理出发,对现有的双视点、多视点和超多视点光场显示技术的成像原理和典型方案进行整理和总结。同时,对现有基于单信源和多信源虚拟视点生成技术进行归纳对比,着重分析各项技术在虚拟视点生成质量和渲染速度等方面的可行性和实现效率,并对光场显示技术的未来发展方向进行了展望。With the advancement of the meta-universe industry,light field display technology has emerged as a research focal point in the realm of information display technology due to its seamless integration of digital and physical worlds.It is exhibiting an upward trend towards more viewpoints,higher viewpoint density,and faster rendering speed.However,extant optical field display technologies are confronted with numerous challenges such as low resolution,limited depth perception,and visual fatigue.Based on the principle of human stereoscopic vision perception,this paper presents a comprehensive overview of the imaging principles and typical schemes employed in current light field display technologies,including double view,multi-view and super-multi-view.Meanwhile,an overview and comparison of current virtual view generation technologies based on single-source and multi-source approaches is provided,with a focus on the feasibility and implementation efficiency of each technology in terms of virtual view quality and rendering speed.Additionally,future development directions for optical field display technology are also anticipated.

关 键 词：近眼显示 光场显示 虚拟视点 神经网络 

分 类 号：TN27[电子电信—物理电子学]