Hough变换的分布式并行计算

Distributed Parallel Computing for the Hough Transform

针对标准Hough变换存储空间需求高、计算量大的缺陷,提出了一种用分布式并行计算实现Hough变换的方法,对分布式算法的分析与设计、Java远程方法调用、多线程同步、负载均衡等关键问题进行了讨论.实验表明,在拥有k台提供资源计算机的分布式系统中,分布式并行计算的时空开销可降低到标准Hough变换的大约1/k.

In order to solve the problem of high memory requirements and computational complexity of standard Hough transform, the paper proposes a pragmatic approach for distributed parallel computing of curve detection using Hough trans- form. In this paper, four key issues are discussed, which are distributed algorithm analysis and design, Java remote method invocation system, multithreading synchronization and load-balancing. The experiments show that the requirement of time and space can be reduced to 1/k by using the proposed algorithm, where k is the number of volunteer-computers in the distributed system.