基于差异图融合和FLICM聚类的SAR图像变化检测  

SAR Image Change Detection Based on Difference Image Fusion and FLICM Clustering

作　　者：董宝兰 汪骏 毕津滔 DONG Baolan;WANG Jun;BI Jintao(School of Information and Engineering,Huangshan University,Huangshan,Anhui 245000,China;School of Mechanical Engineering,Quzhou University,Quzhou,Zhejiang 324000,China)

机构地区：[1]黄山学院信息工程学院,安徽黄山245000 [2]衢州学院机械工程学院,浙江衢州324000

出　　处：《遥感信息》2022年第3期51-56,共6页Remote Sensing Information

基　　金：国家自然科学基金项目(62101206);山东省自然科学基金项目(ZR2019MF014);安徽省创新型省份建设补助资金专项资助项目(2020xzx004);黄山学院校级自然科学研究项目(2020xkjq016)。

摘　　要：针对传统SAR图像变化检测算法易受噪声影响、检测精度低等问题,提出了一种基于差异图融合和FLICM聚类的SAR图像变化检测算法。该算法以对数比和滤波比法生成两幅不同表征的差异图像,并设计融合算子对两差异图的特征信息进行综合,获取高可分性的融合差异图;之后采用局部信息模糊C均值聚类对融合差异图进行处理,分离出其中的变化区域和非变化区域。实验选用漏检数、误检数、正确率、Kappa系数作为评价指标,对算法进行了定性和定量的实验分析。分析结果表明,所提出的算法具有更好的相干斑噪声抑制能力,可有效、准确地检测出变化区域,提高了变化检测的精度。Aiming at the problems that the traditional SAR image change detection algorithm is easy to be affected by noise and low detection accuracy,a SAR image change detection algorithm based on difference image fusion and FLICM clustering is proposed.The algorithm generates two difference images with different representations by logarithm ratio and filter ratio,and designs a fusion operator to synthesize the feature information of the two difference images to obtain a highly separable fusion difference image.Then,fuzzy local information C-means is used to process the fusion difference map to separate the changing region and non changing region.False negative,false positive,precision of correct classification and Kappa coefficient are selected as the evaluation index in the experiment,and the algorithm is analyzed qualitatively and quantitatively.Experimental results show that the proposed algorithm has better speckle noise suppression ability,can effectively and accurately detect the change region,and improve the accuracy of change detection.

关 键 词：SAR图像 差异图 融合算子 变化检测 FLICM聚类 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]