耦合模糊C均值聚类和贝叶斯网络的遥感影像后验概率空间变化向量分析  

A change vector analysis in posterior probability space combined with fuzzy C-means clustering and a Bayesian network

作　　者：李轶鲲 杨洋 杨树文 王子浩 LI Yikun;YANG Yang;YANG Shuwen;WANG Zihao(Faculty of Geomatics,Lanzhou Jiaotong University,Lanzhou 730070,China;National-Local Joint Engineering Research Center of Technologies and Applications for National Geographic State Monitoring,Lanzhou 730070,China;Gansu Provincial Engineering Laboratory for National Geographic State Monitoring,Lanzhou 730070,China)

机构地区：[1]兰州交通大学测绘与地理信息学院,兰州730070 [2]地理国情监测技术应用国家地方联合工程研究中心,兰州730070 [3]甘肃省地理国情监测工程实验室,兰州730070

出　　处：《自然资源遥感》2021年第4期82-88,共7页Remote Sensing for Natural Resources

基　　金：国家自然科学基金项目“基于高分辨率卫星影像的彩钢板建筑与城市空间结构演变关系研究”(编号:41761082);国家重点研发计划(地球观测与导航)项目“星空地遥感立体监测技术”(编号:2017YFB0504201);兰州交通大学(201806)优秀平台共同资助。

摘　　要：在遥感图像变化检测领域中,后验概率空间变化向量分析(change vector analysis in posterior probability space,CVAPS)方法具有诸多优点而被广泛应用。CVAPS法使用支持向量机(support vector machine,SVM)估计后验概率向量,但对中低分辨率遥感影像分类时SVM无法有效处理同物异谱、异物同谱及混合像元问题,从而无法保证最终检测结果的精度。由此,文章针对混合像元问题采用模糊C均值聚类(fuzzy C-means,FCM)进行建模,并耦合简单贝叶斯网络(simple Bayesian network,SBN)以解决混合像元问题及估计后验概率向量,实现了一种新的后验概率空间变化向量分析方法。实验结果表明,本文算法的总体精度和Kappa系数均优于基于SVM的CVAPS算法,算法性能受训练样本的数量影响较小,且参数设置简单,耗时少。文章提出的算法有助于提高遥感图像变化检测的精度和效率。The change vector analysis in posterior probability space(CVAPS)method has been widely used in the change detection of remote sensing images owing to its many advantages.It uses the support vector machine(SVM)to estimate the posterior probability vector.However,in the classification of low and medium resolution remote sensing images,SVM cannot effectively deal with the problems of the same object with the different spectra,different objects with the same spectrum,and mixed pixels and thus cannot guarantee the accuracy of the final detection results.Therefore,this paper adopts the fuzzy c-means(FCM)clustering for modeling and couples the FCM with a simple Bayesian network(SBN)to solve the problem of mixed pixels and estimate the posterior probability vector,thus achieving a new posterior probability space change vector analysis method.The experimental results indicate that,compared to the SVM-based CVAPS algorithm,the algorithm proposed in this study shows higher overall accuracy,higher Kappa coefficient,more reliable performance that is less affected by the number of training samples,simpler parameter setting,and lower time consumption.Therefore,the algorithm proposed in this paper helps to improve the accuracy and efficiency of the change detection of remote sensing images.

关 键 词：遥感图像 变化检测 后验概率空间变化向量分析 模糊C均值聚类 简单贝叶斯网络 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]