虚拟环境下的低延迟协同渲染方法研究
Research on Low Latency Collaborative Rendering in Virtual Environment
王雅妮 1高全力 1王西汉 1毕明洋 1焦子逊1
作者信息
- 1. 西安工程大学 计算机科学学院,陕西 西安 710600
- 折叠
摘要
虚拟现实技术以其独特的沉浸感、交互性,在游戏、直播等各领域的应用愈加广泛.渲染技术是虚拟现实的核心环节之一,而只在本地客户端渲染是目前主要渲染方案,由于实时渲染需要大量的计算资源,导致虚拟现实应用存在响应延迟高、画面刷新率低、图像质量差以及虚拟场景复杂等问题.针对上述问题,该文提出了虚拟环境下低延迟协同渲染方法.首先,将复杂的虚拟环境进行前景交互和背景环境拆分;其次,将渲染工作量较大的背景环境卸载至服务端进行处理;最后,通过运动轨迹预测将渲染的复杂背景以全景图的方式进行分割编码传输至客户端与前景渲染的内容进行合并呈现.经过实验验证,这种协同渲染的方法能够进一步降低传输过程中的CPU/GPU利用率以及网络延迟,提高帧刷新率,进而实现高清低延迟交互.
Abstract
With its unique sense of immersion and interactivity,virtual reality technology is increasingly widely used in various fields such as games and live broadcasting.Rendering technology is one of the core links of virtual reality,and only rendering on the local client is the main rendering scheme at present.Due to the large amount of computing resources required for real-time rendering,virtual reality ap-plications have problems such as high response delay,low image refresh rate,poor image quality and complex virtual scenes.To solve the above problems,we propose a low latency collaborative rendering method in virtual environment.Firstly,the complex virtual environment between foreground interaction and background environment is split.Secondly,the background environment with a high rendering workload is offloaded to the server for processing.Finally,the complex background is segmented and encoded in the way of panorama through motion trajectory prediction and transmitted to the client for merging with the foreground rendered content.After ex-perimental verification,the collaborative rendering method can further reduce the CPU/GPU utilization and network latency in the transmission process,and improve the frame refresh rate,thus achieving high definition and low latency interaction.
关键词
虚拟现实/人机交互/协同渲染/全景图传输/高清低延迟Key words
virtual reality/human-computer interaction/collaborative rendering/panorama transmission/high definition and low latency引用本文复制引用
基金项目
国家自然科学基金(61902300)
陕西省重点产业创新链项目(2020ZDLGY07-05)
陕西省教育统计数据研究中心专项研究课题(21JTY010)
出版年
2023
国家科技期刊平台
NSTL
维普
万方数据