一种改进的随机Hough变换检测圆的方法

An improved Method of the Randomized Hough Transforms to Detect the Circle

针对随机Hough变换检测圆准确率低的问题,提出了一种改进的随机Hough变换算法.算法利用原标准随机Hough变换的检测结果,确定出合适的阈值,将检测结果中平均误差超过此阈值的圆视为错检圆,并将这些圆上的点取出,重新进行随机Hough变换,直到正确检测出所有圆.模拟数据和真实图像的数值实验都表明改进算法的检测准确率明显提高.

For the problem of low accuracy to detect circle by the randomized Hough transform, this paper puts forward an improved algorithm based on the randomized Hough transform. Firstly the algorithm uses the results of the original randomized Hough transform, determines the proper threshold. Secondly the circles which the average error surpasses the threshold is served as fault circles, and these points on the circles are removed, and done the randomized Hough transform again until all circles are detected correctly. The experiments of numerical data and real images show that the detection accuracy of the algorithm greatly improves.