基于边缘检测和自动种子区域生长的图像分割算法

针对传统区域生长算法对噪声敏感和初始种子过度依赖的问题,提出一种基于边缘提取和自动区域生长相结合的分割算法,即先利用Canny算子对图像进行边缘提取,得到边缘像素点集,再利用灰度直方图从点集中选取自动种子点,结合平均相似度,采用区域生长法进行图像分割。该算法能够实现阈值和种子的自动选取,在传统算法中很难实现。实验结果表明该算法有效。

An image segmentation algorithm based on edge extraction with automatic seeded region growing

In view of the problem that traditional region growing algorithm is noise-sensitive and over-reliance on the initial seed,a segmentation algorithm based on edge extraction and automatic regional growth is proposed.After the set of edge pixels is extracted out by using Canny operator,the seeds are automatically selected from it according to the histogram.With the average similarity is considered,a new region growing method can be used for image segmentation.Experimental results show that the new algorithm is effective.