基于模糊最大似然估计算法的遥感影像分割  

Fuzzy Maximum Like Hood Estimation Based on Remote Sensing Image Segmentation

作　　者：徐艳[1] XU Yan(School of Geomatics,Liaoning Technical University,Fuxin 123000,China)

机构地区：[1]辽宁工程技术大学测绘与地理科学学院,辽宁阜新123000

出　　处：《测绘与空间地理信息》2019年第1期163-166,170,共5页Geomatics & Spatial Information Technology

摘　　要：针对模糊K-均值算法依赖于群集原型的初始估计和对于数据中所存在的子群数目做出假设的缺点,结合最大似然估计,提出了不依赖先验假设的模糊聚类法——基于模糊最大似然估计的遥感影像分割算法。该算法在模糊最大似然估计算法中用模糊协方差来计算后验概率,用后验概率矩阵代替隶属度矩阵来进行划分。先用模糊K-均值进行图像预处理,然后用模糊最大似然估计算法进行分割。此外,本文用性能指标参数——超体积指标FHV来评价最优的类别数目。本文通过对模拟影像和真实影像的实验,验证了该算法的有效性和准确性。The K-means clustering algorithm and the fuzzy K-means algorithm were told in detail.The fuzzy K-means algorithm depends on initial guesses of cluster prototypes,and on assumptions made as to the number of subgroups present in the data.Then on this base,this study is combined with the maximum likelihood estimation,reports on a method for carrying out fuzzy classification without a priori assumptions on the number of clusters in the data set,called fuzzy maximum likelihood estimation based on remote sensing image segmentation.In the fuzzy maximum likelihood estimation,the posteriori probability is calculated by fuzzy covariance matrix and the posteriori probability matrix is used to replace the membership matrix to divide images.First,the fuzzy K-means is to be used,and then the fuzzy maximum likelihood estimation is used for segmentation.In addition,this paper tries to use performance measures--hyper volume measure(FHV)to evaluate the optimal number of classes.It has been tested on a simulated images and real images.

关 键 词：遥感影像分割 模糊K-均值算法 模糊最大似然估计 性能指标参数 超体积函数 

分 类 号：P237[天文地球—摄影测量与遥感]