基于期望最大理论的无监督图像分割  

Expectation-Maximization Algorithm-Based Unsupervised Image Segmentation

作　　者：高敬惠[1] 李玉海[1] 刘国丽[1] 

机构地区：[1]河北工业大学廊坊校区计算机系

出　　处：《微计算机信息》2007年第24期309-310,212,共3页Control & Automation

基　　金：河北省教育厅科学研究项目(2004416)基于小波分析的视频图像压缩算法研究

摘　　要：本文提出了一种新的基于期望最大化以及贝叶斯信息准则的图像分割方法。首先,运用K均值方法初始化图像分布,运用期望最大算法估计输入图像参数数据,且图像中类的数目由贝叶斯消息准则自动确定。运用最大似然标准将像素归类于最相近的类中。本法的优点在于不过分依赖于原始估计,可以用来进行无监督的图像的分割。运用两幅真实图像进行了实验,结果表明此方法有效。Tbe paper proposes a novel image segmentation method based on Expectation-Maximization and Bayesian Information Criterion. The Expectation-Maximization theory is used to estimate the data distribution of the input image firstly. The number of class is calculated by Bayesian Information Criterion. The Maximum Likelihood is employed to classify the image pixels into the nearest class. The excellence of the proposed method is being independent to original estimate and it can be used in unsupervised image segmentation, The synthetical and real images are used in the experiment. The results show that this method is efficient.

关 键 词：期望最大 贝叶斯信息准则 图像分割 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]