应用混沌多目标规划理论融合的图像分割  

Image Segmentation Based on Combining Chaos and Multiobjective Programming Theory

作　　者：张新明[1] 郑延斌[1] 张慧云[1] 

机构地区：[1]河南师范大学计算机与信息技术学院,河南新乡453007

出　　处：《小型微型计算机系统》2010年第7期1416-1420,共5页Journal of Chinese Computer Systems

基　　金：河南省重点科技攻关项目(092102210017)资助;河南省教育厅科技攻关项目(2007520024;2008B520021)资助

摘　　要：针对二维最大熵和二维最大类间熵阈值化方法通用性不强,使得他们在某些分割应用场合失效的问题,提出应用混沌多目标规划理论融合的图像分割方法.这种新方法利用多目标规划理论将前两种方法有机结合,得到既满足二维最大熵原则,又满足二维最大类间交叉熵原则的最佳阈值,并且将混沌优化算法和递推算法相结合,使计算复杂度大大降低.仿真实验结果表明,本文提出的方法好于单独使用一种方法的分割结果,而且有较好的普适性.The thresholding method based on 2-D maximum entropy and the one based on 2-D maximum inter-class cross entropy do not work well for a lot of images,so a new image segmentation method based on combining chaos and multiobjective programming theory is presented in this paper,which combines 2-D maximum entropy and 2-D maximum inter-class cross entropy using multi-objective programming theory so that the optimal threshold value can be obtained which can satisfy both the former and the latter,and which combines chaos optimization algorithm and recurring to reduce computational complexity.Experimental results show that the proposed method＇s segmentation effect is not only better than that of the two maximum entropy methods each independently,but also this method is more universal.

关 键 词：图像分割 二维最大类间交叉熵 二维最大熵 阈值化 多目标规划 混沌优化 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]