Hough变换和区域分离-合并相结合的分割算法

针对传统区域分离-合并算法中过分割的问题,提出一种基于边缘提取和区域分离-合并相结合的分割算法。即利用Canny算子对图像进行边缘提取,得到边缘像素集。然后将像素集作为边缘线,采用区域分离-合并算法进行图像分割。由于Canny算子所提取的像素集是不完整的,所以需要用Hough变换对其进一步完善,得到连续的边缘线。该算法能够提高区域分离-合并算法的精确度,较好地解决其过分割的问题。实验结果表明,该算法有效。

A segmentation method based on the combination of Hough edge and region split-merger algorithm

In order to solve the over-segmentation of the traditional regional split-merger algorithm,a segmentation method based on the combination of Hough edge detection and region split-merger algorithm is proposed.By using the Canny operator to extract image edge,a set of edge pixels is got and treated as edge line.The images are segmented with region split-merger algorithm.Because the set of pixels extracted by the Canny operator is not complete,it needs to be improved further through using the Hough transformation,and then the continuous edge line is got.This algorithm can increase the accuracy of region split-merger algorithm,and better solve the problem of over segmentation.The experimental results show that this algorithm is effective.