华中科技大学学报（自然科学版）2011，Vol.39Issue(z2)：85-87,102,4.
鲁棒性好的彩色图像分割方法
Robust segmentation method to illumination variation for color image
摘要
Abstract
A robust hybrid segmentation method was proposed which for color image by combination of graph theory clustering and region growing techniques. The method uses region growing to finish initial segmentation, the final segmentation images are produced by minimum spanning tree(MST) method which looks a region as a node, and the best thresholding is obtained by objective function of maximum weighted entropy. The shadow and color apperceive produced by illumination change are weakened. Experiment results show that the new algorithm ensures the color image excellent segmentation and faster segmentation speed.关键词
彩色图像分割/区域生长/图论/最大加权熵/鲁棒性Key words
color image segmentation/ region growing/ graph theory/ maximum weighted entropy/ robustness分类
信息技术与安全科学引用本文复制引用
张学习,谢云,杨宜民,夏益民..鲁棒性好的彩色图像分割方法[J].华中科技大学学报（自然科学版）,2011,39(z2):85-87,102,4.基金项目
广东省科技计划资助项目(2009B010900030). （2009B010900030）
