基于二维灰度熵及混沌粒子群的图像阈值选取  

Image Thresholding Based on 2-Dimensional Gray Entropy and Chaotic Particle Swarm Algorithm

作　　者：吴一全[1,2] 周怀春[2] 纪守新[1] 张晓杰[1] 

机构地区：[1]南京航空航天大学电子信息工程学院,江苏南京210016 [2]华中科技大学煤燃烧国家重点实验室,湖北武汉430074

出　　处：《北京理工大学学报》2011年第12期1428-1434,共7页Transactions of Beijing Institute of Technology

基　　金：国家自然科学基金资助项目(60872065);煤燃烧国家重点实验室(华中科技大学)开放基金重点资助项目(FSKLCC1001);计算机软件新技术国家重点实验室(南京大学)开放基金资助项目(KFKT2010B17)

摘　　要：为了同时考虑直方图的概率信息和类内灰度级的均匀性,提出了基于灰度级-梯度二维直方图的Shannon灰度熵及Tsallis灰度熵阈值选取方法.给出了Shannon灰度熵和Tsallis灰度熵的定义及其一维阈值选取方法,导出了二维Shannon灰度熵及Tsallis灰度熵阈值选取公式及其快速递推算法,并利用混沌粒子群算法寻求两种阈值选取方法的最佳阈值.实验结果表明,与基于改进的二维最大熵及粒子群递推的阈值选取方法相比,所提出方法的分割图像能更准确地反映原始图像的边缘、纹理及细节信息.To consider simultaneously the histogram probability information and the uniformity of within-cluster gray level in the 2-dimensional maximum entropy thresholding method, the 2-dimensional Shannon gray entropy and Tsallis gray entropy thresholding methods are proposed based on gray level-gradient histogram in this article. First, the Shannon gray entropy and Tsallis gray entropy were defined and the one-dimensional thresholding methods were given. Then 2-dimensional Shannon gray entropy and Tsallis gray entropy thresholding formulae and their fast recursive algorithms were derived, and the chaotic particle swarm optimization algorithm was used to search the best thresholds. Lots of experiments were done and the results show that, compared with the thresholding method based on improved 2-dimensional maximum entropy and particle swarm optimization, the obtained segmented images using suggested method can reflect the edge, texture and details of the original images with more accuracy.

关 键 词：图像阈值选取 二维Shannon灰度熵 二维Tsallis灰度熵 快速递推算法 混沌粒子群优化 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]