GPU加速在集成成像的应用

GPU Acceleration Used in Integral Imaging

针对在计算3D集成成像过程中耗时较多的问题,采用了一种新的硬件加速方法——GPU加速,分别在不同透镜数目和三维物体复杂度下,对CPU和GPU的运算时间进行对比。结果表明:计算复杂度大于数据拷贝时间时,GPU的整体加速效果明显,并且随着计算复杂度的提高,加速效果越来越显著。

The calculation of Three-Dimensional（3D）integrated imaging usually consumes a lot of time. In order to solve this problem, a new hardware acceleration method, GPU acceleration, is introduced in this paper. This paper compares CPU＇s and GPU＇s computing time in different lens number and complexity of the three-dimensional object. The result shows that the effect of GPU＇s acceleration is apparent when the computational complexity is greater than the data copying time. The effect can be more apparent when the complexity rising.