一种快速的动态Hough变换及其并行实现

A Fast Dynamic Hough Transform and Its Parallel Implementation

本文提出了用于直线检测的快速动态Hough变换.它采用一维Hough聚类技术,因而其参数空间的峰值检测和解释比较简单,线段端点的确定及其连通性的检验也很容易,并节省了大量存储空间.此外,它只需进行整数运算,且运算次数只与象素点数呈线性关系,因而其速度很快.实验结果表明,它可以准确、快速地检测出直线.为满足实时应用的需要,本文还给出了该变换在Transputer网络上的并行实现方案.

A fast dynamic Hough transform which is used for line detection is proposed. It uses the ID Hough clustering technique, so the peak detection and interpretation of the parameter space are very easy, and the endpoints and the connectivity of the line segments can be determined conveniently. A large amount of memory space can also be saved. Moreover, only integer operations are required in this transform, and the number of operations proportional to the number of edge points, so that transform can be carried out very fast. In the last, the parallel implementation of the transform based on a transputer network is also given to meet the needs of real time application.