您的位置:首页  > 论文页面

基于反向光线跟踪技术和感兴趣区域的超多视点图生成加速算法

发表时间:2021-01-08  浏览量:103  下载量:13
全部作者: 曾晗琦,邢树军,桑新柱
作者单位: 北京邮电大学信息光子学与光通信国家重点实验室;清华大学精密测试技术及仪器国家重点实验室
摘 要: 对于包含运动物体的复杂场景而言,使用计算机生成用于三维显示的超多视点图的效率往往较低。为解决该问题,本文提出了一种基于反向光线跟踪技术和感兴趣区域(region of interest,ROI)的生成加速算法,在保证显示质量的前提下对超多视点图的生成过程进行加速。在每一帧中,内容变化的部分被检测并标记为ROI,只有该部分的像素被重新渲染。实验结果表明,内容生成过程的速率可被提升至原来的2~9倍。此外,在分辨率为3 840×2 160的复杂应用情况下,输出帧率由3.5帧/秒提升至超过30帧/秒,满足了实时交互的需求。
关 键 词: 电子、通信与自动控制技术其他学科;自由立体显示;光线跟踪;超多视点成像;三维显示
Title: An accelerating algorithm for generation of super-multiview images based on backward ray-tracing technique and region of interest
Author: ZENG Hanqi, XING Shujun, SANG Xinzhu
Organization: State Key Laboratory of Information Photonics and Optical Communications, Beijing University of Posts and Telecommunications; State Key Laboratory of Precision Measurement Technology and Instruments, Tsinghua University
Abstract: To overcome the limitation that computer generated super-multiview images for three-dimensional (3D) display is not efficient enough for complex scenes with moving object, an accelerating algorithm based on the backward ray-tracing technique and the region of interest (ROI) is presented, so that the generating process of the super-multiview images can be accelerated while the image quality is maintained. For each frame, the changed part of the scene is detected and marked as the ROI, and only the pixels located in the ROI are re-rendered. Experimental results show that the generating process can be accelerated to be 2-9 times faster. In addition, the output frame rate (frames pet second) is improved from 3.5 to over 30 for the complex application situation with a resolution of 3 840×2 160, which meets the real-time demand.
Key words: other subjects of electrics, communication and autocontrol technology; auto-stereoscopic display; ray-tracing; super-multiview image; three-dimensional display
发表期数: 2020年12月第4期
引用格式: 曾晗琦,邢树军,桑新柱. 基于反向光线跟踪技术和感兴趣区域的超多视点图生成加速算法[J]. 中国科技论文在线精品论文,2020,13(4):504-511.
 
3 评论数 0
暂无评论
友情链接