改进的随机Hough变换圆检测算法

Improved randomized Hough transform algorithm for circle detection

针对随机Hough变换会产生大量无效累积的问题,提出了一种改进的随机Hough变换算法来检测圆,该算法利用梯度来预先判断随机采样的三个点是否在同一个圆上,从而大大减少了无效累积;另外,该算法还在圆参数的计算、阈值的确定、候选圆的确认等方面进行了改进。实验结果表明,该算法精度高,速度快,检测性能有了较大提高。

To solve the problem of lots of invalid accumulation yielded by randomized Hough transform, an improved randomized Hough transform algorithm was proposed to detect circle. It used gradient beforehand to determine whether the three points sampled randomly were on a same circle, so it could reduce invalid accumulation greatly. Additionally, it had improvements in calculating the circle parameters, determining the threshold, and confirming the candidate circle. The experimental results indicate that the algorithm has high precision and high speed, and its capability of detecting circle is enhanced greatly.