摘要
为了解决图像处理中计算量大、处理速度慢的问题,提出了一种基于流处理器的图像并行处理系统。该系统采用Strom-1系列SP16HP-G220流数字信号处理器构建硬件平台,根据流处理器体系结构特点建立了流程序模型,并参考图像处理实际需求选取具有代表性的灰度变换与Sobel算子进行算法并行性分析与算法流化,利用流化过的算法与传统算法进行了对比实验,实验结果表明,灰度变换加速比为199,Sobel算子边缘检测加速比为175,满足机器人系统对图像处理实时性的需求。
For solving the existing problem of data-intensive and slow speed in image processing,a parallel work system for image processing is proposed based on stream processor.The hardware platform employing SP16HP-G220 DSP of Strom-1 series is developed.The program model of stream processor by architectural characteristic of stream processor is established,and in actual application of image processing typical grey-scale convertion and Sobel operator that execute the analysis of parallel algorithm and algorithm of stream-oriented are selected.The comparative experiment of stream-oriented and traditional algorithm are carried out.The experimental results show that the speedup of grey-scale intensity image can reach 199,and the speedup of Sobel operator can reach 175.This can meet the real-time requirement for image processing in robot system.
出处
《电子器件》
CAS
2010年第6期756-761,共6页
Chinese Journal of Electron Devices
基金
云南省应用基础研究基金资助(2009ZC041M)
昆明理工大学引进人才培养基金资助(2010-07)