摘要
为了提高计算机视觉中图像处理的速度,深入分析了计算机视觉中图像处理的三个层次的并行计算特征.以数据并行的处理元阵列芯片为基础,通过对其进行不同组合得到高性能嵌入式图像处理系统.该系统为图像处理提供了不同层次的数据并行性和任务并行性,满足了图像处理对并行计算的需求,为实时嵌入式图像处理提供了较高的计算性能.此外,处理元阵列芯片的实现方式又保证了其具有较小的体积,满足了嵌入性的要求.
In order to improve the speed of image processing in computer vision, the three levels parallel computing characteristics are deeply analyzed. Using data parallel processing element array composes different high performance embedded image processing systems. These kinds of systems can provide different levels data parallelism and task paralldism which can match image processing＇ s parallelism. Thus theses kinds of systems can speedup the image processing applications for real time embedded image processing. Furthermore, the realization of processing element chip enable it has a small volume which can match the requirement of embedded applications.
出处
《微电子学与计算机》
CSCD
北大核心
2008年第6期38-41,共4页
Microelectronics & Computer
基金
陕西省自然科学基础研究计划项目(2007F11)
中国博士后科学基金项目(20070421101)
关键词
嵌入式处理器
并行计算
处理元阵列
图像处理器
embedded processor
parallel computing
array processing element
image processor
