GPU加速的傅里叶变换轮廓术并行计算方法  

GPU Accelerated Parallel Computing Method for Fourier Transform Profilometry

作　　者：赵小敏[1] 周波[1] 刘春媛[1] 陶金[1] 

机构地区：[1]黑龙江科技学院,黑龙江哈尔滨150027

出　　处：《机械制造与自动化》2013年第2期141-144,共4页Machine Building & Automation

基　　金：黑龙江省教育厅科学技术研究项目(12511474);黑龙江省研究生创新科研项目(YJSCX2011-007HKJ)

摘　　要：傅里叶变换轮廓术需要在高分辨率图像上进行相位计算,其耗时较长,不能满足实时处理的要求。相位计算时其待处理的和已处理的数据都是相对独立的图像像素,且计算密度极大,适于并行计算。因此,利用图形处理器的多线程并行处理能力,在GPU上实现了相位的并行计算,解决了在CPU上相位计算速度较慢的问题。实验表明在相位计算质量相同的情况下,经过GPU加速获得了相对于CPU一到两个数量级的加速比,为将傅里叶变换轮廓术应用于实时三维测量奠定了坚实的基础。The phase computation on the high-resolution images in the Fourier transform profilometry takes long time and does not meet the real-time processing requirement. The phase computation is not dependence on the neighborhood pixels with high arithme- tic intensity. It is suitable for parallel. Therefore, it is possible to make the phase computation implementation on GPU using the multi- threading parallel processing, and solve the low efficiency on CPU. Experiments show that it obtains the speedup magnitude of one to two orders by GPU acceleration with the same quality. So it lays the foundation for the Fourier transform profilometry applyied to the real-time three-dimensional measurement.

关 键 词：图形处理器加速 并行计算 傅里叶变换轮廓术 统一计算设备架构 计算相位 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]