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

作　　者：杨家红[1] 刘杰[1] 钟坚成[1] 何明志[1] 

机构地区：[1]湖南师范大学电子与信息工程系,长沙410081

出　　处：《中国图象图形学报》2010年第1期63-68,共6页Journal of Image and Graphics

基　　金：湖南省自然科学基金项目(08JJ313);国家烟草专卖总局科技开发项目(2006371)

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,该算法简单、快速,且能得到较其他算法更好的分割效果。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.

关 键 词：分水岭 自动种子区域生长 小图像区域消解 各向异性扩散 彩色图像 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]