真三维活动视频数据的优化研究

提出了一种基于点阵的真三维视频显示技术,该系统利用LED为单元节点组成三维空间阵列,用于显示真三维活动影像。由于数据量巨大,为了加快处理速度,利用CUDA编程模型对计算过程进行优化,把处理过程中可以并行计算的部分交由GPU执行。先把要处理的视频数据传到内存中,由CPU进行一些预处理,然后传到显存,由GPU对视频运动过程等进行处理,处理完后再传到内存,由CPU进行一些后续处理,最终把处理后的数据传出加以显示或存储。通过比较仅由CPU处理与用GPU优化后的计算时间,发现优化后计算速度比优化前快了几十到几百倍,而且数据量越大,优化效果越好,核心多的GPU所得到的加速比大,最后在实验部分给出了用OpenGL仿真的结果。

Research on optimization of true 3-D video data

A true 3-D video display technology based on point array is presented.The system consists of three dimensional space array by taking LED as a unit node,which displays true 3-D active images.Due to the large amounts of data,CUDA programming model is adopted to optimize the calculation process,in which some parts that can be calculated parallelly are executed by GPU.First,the video data under processing is sent to the memory for preprocessing by CPU,and then sent to the local video memory to process the video movement course by GPU.After that,it is sent to the memory again for posttreatment by CPU.Finally,the processed data is sent out for display or storage.After comparing the computation time on GPU and CPU,it is found that the calculation speed of the optimized algorithm is increased by hundreds of times,and the more data,the effect is better.The results simulated by OpenGL are given.