结合分水岭与自动种子区域生长的彩色图像分割算法

杨家红; 刘  杰; 钟坚成; 何明志

doi:10.11834/jig.20100111

图像分析和识别 | 浏览量 : 0 下载量: 669 CSCD: 0

PDF
导出
分享
收藏
专辑

结合分水岭与自动种子区域生长的彩色图像分割算法
A Color Image Segmentation Algorithm by Integrating Watershed with Automatic Seeded Region Growing
2010年15卷第1期页码：63
纸质出版：2010
DOI： 10.11834/jig.20100111
稿件说明：

移动端阅览

杨家红, 刘杰, 钟坚成, 何明志. 结合分水岭与自动种子区域生长的彩色图像分割算法[J]. 中国图象图形学报, 2010,15(1):63. DOI： 10.11834/jig.20100111.

YANG Jia-hong, LIU Jie, ZHONG Jian-cheng, HE Ming-zhi. A Color Image Segmentation Algorithm by Integrating Watershed with Automatic Seeded Region Growing[J]. Journal of Image and Graphics, 2010, 15(1): 63. DOI： 10.11834/jig.20100111.

摘要

为了对彩色图像进行有效分割，提出了一种将分水岭算法与自动种子区域生长相结合的分割算法。该算法首先利用基于边缘信赖度的各向异性扩散方程对图像进行平滑预处理，以便在去除图像噪声的同时，保持图像的边缘信息，从而有效解决了分水岭算法的过分割现象；然后在此基础上，设计了一种基于色调均值差的自动种子区域选取算法，即根据色调均值和饱和度均值的相似性来对分水岭算法分割形成的区域进行种子区域生长，并利用小图像区域消解算法对区域生长所遗留的小区域进行消解，减少了图像中小区域的数目，从而实现了对彩色图像的有效分割。实验结果表明

Abstract

A color image segmentation algorithm by integrating watershed with automatic seeded region growing is proposed in this paper Anisotropic diffusion based on edge confidence is used for image smoothing preprocessing

which can eliminate noise

while maintain image edge information

and so that the watersheds over segmentation problem is effectively solved The algorithm of automatic seeded region selection based on the mean hue difference is designed to make seeded region growing based on the regions similarity of mean hue value and mean saturation value Furthermore

an algorithm of small region removing is designed to removing the small regionAs a result

the color image segmentation is efficiently realized The experimental results demonstrate that this algorithm is simple and fast

can obtain better segmentation results compared with other algorithms