一种基于广义Hough变换思想的多圆检测方法

A Method for Multiple Circle Detection Based on Generalized Hough Transform

根据广义Hough变换(GHT)统计聚类思想和圆的定义、特点,提出了一种针对多个圆的检测方法。该方法在无任何先验知识(如目标圆圆心的大概位置,半径的精确范围)的情况下,能快速检测出图像中的多个圆。实验结果表明,该方法效果较好,而且对椒盐噪声、直线、椭圆等非圆物体有很好的抗干扰性。

This paper proposes a method for multiple circle detection based on Generalized Hough transform＇ statistic and clustering principle and circle＇ definition and characteristics. Circles will be extracted quickly from a complex image without a prior knowledge （e.g. center and radius＇ narrow range of circle which will be detected）. The experimental results demonstrate that the proposed method performs well even against non-circle noise such as salts, lines,ellipses.