基于边缘检测与分裂合并的图像分割算法

An image segmentation algorithm based on edge detection and split-merge

针对传统分裂合并算法容易产生方块效应与过分割的缺点,提出了一种结合边缘检测和分裂合并的图像分割算法。该算法直接利用图像的边缘信息进行分裂,不断将图像分裂为一些不规则形状的一致性区域,然后根据一定规则将相似的区域合并。实验表明,该算法能大幅减少分裂次数,并有效克服方块效应和过分割等缺点,图像分割效果较好。

An adaptive image segmentation algorithm based on edge detection and split-merge is presented in this paper in order to solve the problem that tradition split-merge image segmentation algorithm can easily lead to over-blocking effects and over- segmentation. The algorithm directly uses the edge information during the split process by continuously splitting image into the con- sistency of irregular shapes, then combines the regional shape according to certain rules. Experimental results showed that the pro-posed algorithm can significantly reduce the number of division and effectively overcome the box effects and over-segmentation, then obtains good effect in image segmentation.