基于空间模糊C均值聚类和贝叶斯网络的抗噪声遥感图像变化检测  

Noise-resistant change detection for remote sensing images based on spatial fuzzy C-means clustering and a Bayesian network

作　　者：王子浩 李轶鲲 李小军 杨树文 WANG Zihao;LI Yikun;LI Xiaojun;YANG Shuwen(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》2023年第4期96-104,共9页Remote Sensing for Natural Resources

基　　金：国家自然科学基金项目“基于脉冲耦合神经网络的高光谱遥感图像融合方法研究”(编号:41861055);中国博士后基金项目(编号:2019M653795);兰州交通大学优秀平台(编号:201806)共同资助。

摘　　要：目前,大部分遥感变化检测算法无法有效处理受高斯、椒盐和混合噪声污染的图像。为了解决这一问题,文章将能够在噪声污染条件下有效分解混合像元的5种基于邻域空间信息的模糊C均值聚类(FCM_S1,FCM_S2,KFCM_S1,KFCM_S2和FLICM)算法分别与简单贝叶斯网络(simple Bayesian network,SBN)相结合,在后验概率空间变化向量分析(change vector analysis in posterior probability space,CVAPS)框架下,实现了5种能够较好地抗高斯、椒盐和混合噪声的遥感变化检测方法。对比实验证明,该文所提出的变化检测算法对高斯、椒盐和混合噪声具有较好的鲁棒性。Currently,most change detection algorithms for remote sensing images fail to effectively process images polluted by Gaussian,impulse,or mixed noise.To address this problem,this study presented five fuzzy C-means(FCM)clustering algorithms(FCM_S1,FCM_S2,KFCM_S1,KFCM_S2,and FLICM)based on neighborhood space information.These algorithms,which can efficiently decompose mixed pixels in the presence of noise pollution,were combined with a simple Bayesian network(SBN).Under the framework of change vector analysis in posterior probability space(CVAPS),this study developed five change detection methods for remote sensing images,exhibiting high resistance to Gaussian,impulse,and mixed noise.Comparative experiments demonstrate that the change detection algorithms proposed in this study manifest high robustness against the above-mentioned noise.

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

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]