摘要
为了解决全景视频生成系统中的速度瓶颈,讨论了一种基于显卡硬件加速的全景视频生成方法,该方法将显卡中的图形处理器(graphic processing unit,GPU)作为CPU的协处理器,充分发挥GPU强大的图形处理能力,协助CPU处理那些运算量巨大的图像处理任务,如白平衡和透视变换处理等,从而将CPU从繁重的图像处理工作中解放出来,并在很大程度上提高全景视频的生成效率,使之达到实时的生成效果。这一GPU协同处理的全景视频生成架构已在一台全景摄像机系统样机上实现,并能够实时地生成360°的全景视频,表明该全景视频生成架构在工程上具有可行性。
The purpose of this paper is to resolve the speed bottleneck in panoramic video generating system. This paper introduces a panoramic video generating method based on video card hardware acceleration, which takes GPU as CPU＇s coprocessor. By this method, users can take full advantage of the powerful graphic processing ability of GPU and make it cooperate with CPU to deal with those complicated graphic processing task, such as white balance and perspective trans-formation, and release CPU from heavy work of graphic processing. This method can greatly improve the panoramic video generating efficiency and achieve a real time effect when generating panoramic video. This architecture of GPU assisted processing has come true in a panoramic camera with the capability to create a panoramic video of 360 degree around in a real time, indicating that this architecture is viable on project.
出处
《国外电子测量技术》
2009年第5期78-82,共5页
Foreign Electronic Measurement Technology
关键词
全景视频
图形处理器
GPU通用计算
全景视频生成架构
panoramic video
graphic processing unit
general purpose computation on the GPU
panoramic video generating architecture
