摘要
图象的多解析度处理技术广泛地用于计算机视觉研究中.然而这种图象处理技术要求采用功能强大的计算机,如由多个 Transputer 处理器组成的并行计算机,来处理繁重的运算任务.本文介绍如何在 Transputer阵列机上实现用于产生几种图象金字塔数据结构的并行算法.实验结果表明,在并行 Transputer 计算机上运行上述算法的程序,运行速度比个人计算机完成同样任务时的运行速度提高数倍.
Image multiresolution processing is an ingenious technique commonly used in computer vision.However,powerful computers such as multitransputer machine must be adopted to deal with the enormous computation needed in this technique.This paper describes how to use a transputer array to implement parallel algorithms for the generation of several image pyramid data structures. Experiments show that the parallel programs behave with much better perfor- mances than an IBM personal computer.
出处
《北方交通大学学报》
CSCD
北大核心
1993年第A08期16-158,共1页
Journal of Northern Jiaotong University
基金
国家教委博士基金资助项目
关键词
计算机视觉
图象处理
并行处理
computer vision
multiresolution image processing
transputer
parallel processing
