用点Hough变换实现圆检测的方法

Circle Recognition Through a Point Hough Transformation

给出了一种新的Hough 变换圆检测方法——点Hough变换（PHT）。该方法根据圆 周上任意两条不平行弦的中垂线相交与圆心的几何性质，同时选取曲线上3点进行计算，将 传统Hough变换圆检测时的三维参量统计变成一维参量统计，极大地降低了计算复杂性和对 资源的需求。为了克服任意选取组合点可能带来的计算量增加及统计结果的分散程度提高, 文中提出了点的选择方法。合成图和实际图像的实验结果表明，该方法用于普通图像中圆检 测时快速、稳定、准确。

This paper introduces a new method, point hough transform (PHT), to detect a circle in image. The PHT technique makes use of the property that the i ntersection point of two lines that perpendicularly bisect segments of a circle is the circle's center. Every group of three points on a circle that do not belo ng to one line is correspond to a point in variables space. In variables space, highly voted point provides an indication of the existence of a circle. Compared with existing methods, the PHT is faster and more economic.