沉浸式全景视频芯片基础问题和关键技术研究

沉浸式虚拟现实技术是当前最热门的研究课题和科技热点，各大厂商纷纷推出各类眼镜、头盔设备用于虚拟现实体验。当前沉浸式虚拟现实技术发展面临的最大问题是全景视频内容的缺失，其中最主要的原因是全景视频采集芯片仍处于空白状态，芯片研制面临重大技术瓶颈。本课题旨在面向沉浸式虚拟现实应用，研究芯片级的实时全景视频拼接、全景视频数据结构、数据压缩传输、全景场景还原等。着重解决海量视频数据下的芯片IO带宽优化、多层次存储与片内数据复用、球面均匀像素采样、超高性能图像拼接算法与硬件、H.265全景视频压缩、FPGA原型系统开发、新型人机交互模式等关键问题。最终实现更具沉浸感、交互感和刺激感的视频体验。

Currently, Immersive Virtual Reality is the hottest research topic and technology hotspot. Many immersive virtual reality display devices are released by famous high-tech company. The main bottleneck of Immersive Virtual Reality is that the missing of panoramic video content. Due to the technology limit, the panoramic video capture IC is still blank by now. This project aims to do research on chip level panoramic video stitch, panoramic video data modeling, panoramic video coding and transmission, and the immersive video scene reconstruction for a video based immersive virtual reality system. We aim to solve the key technology challenges like chip IO optimization under massive video data, on-chip multi-level memory architecture and data reuse, uniform sphere pixel sampling, high-performance video stitch algorithm and hardware, H.265 video encoder for panoramic video, FPGA demo system, next generation of human-computer interaction, and so on. Finally, a more immersive, interactive, and imaginative video system will be achieved in this project.