基于差异图和块分类的雷达图像变化检测  

Change Detection in SAR Images Based on Difference Image and Block Classification

作　　者：王平 WANG Ping(College of Mathematics and Computer,Wuyi University,Wuyishan 354300,China;Key Laboratory of Cognitive Computing and Intelligent Information Processing of Fujian Education Institutions,Wuyishan 354300,China)

机构地区：[1]武夷学院数学与计算机学院,福建武夷山354300 [2]认知计算与智能信息处理福建省高校重点实验室,福建武夷山354300

出　　处：《长春师范大学学报》2020年第8期42-49,共8页Journal of Changchun Normal University

基　　金：福建省自然科学基金面上项目“基于迁移学习的小样本茶叶病虫害自动识别技术研究”(2019J01833);福建省教育厅科研基金项目“基于效用网络资源分配及QoS约束路由算法的研究”(JA15522);认知计算与智能信息处理福建省高校重点实验室开放基金项目“基于增量学习模式的多时相合成孔径雷达图像变化检测方法研究”(KLCCIIP2019203);福建省本科高校教育教学改革研究项目“基于数学建模创新能力培养的运筹学课程教学改革研究”(FBJG20190281);武夷学院科研基金项目“无线传感器网络的QoS体系结构研究”(xl201016)。

摘　　要：针对两时相合成孔径雷达图像的变化检测中直接用比值法生成差异图存在的不足,提出了采用像素级和特征级信息互补融合的处理方法。根据像素级强度比值和特征级邻域块强度比值互补融合构建差异图,然后对差异图像素利用层次模糊C均值聚类的方法得到高概率变化类、不变化类以及中间不确定类,进而以这些变化类和不变化类像素点为中心构建两输入时相图的邻域图像块作为分类器的样本集,且为进一步提高分类器的精度,采用基于遗传算法优化的极限学习机进行分类,将聚类结果和分类器结果相结合,生成两时相的合成孔径雷达图像的变化检测图。对多个真实的合成孔径雷达图像数据进行测试,表明该方法对噪声的影响具有较高的鲁棒性,能够有效地检测两时相合成孔径雷达图像的变化情况,且具有较高的正确率和KAPPA系数值。In order to overcome the shortcomings of generating difference image directly by ratio method in the change detection of two-phase SAR images,a new processing method using pixel-level and feature-level information complementary fusion is proposed.The difference image is constructed by complementary fusion of pixel-level intensity ratio and feature-level neighborhood block intensity ratio,and then the pixel of the difference image is processing by using the hierarchical fuzzy C-means clustering method to get high probability the change class,the non-change class and intermediate uncertain class.Then,the neighborhood image block of the two-input phase diagram is constructed as the sample set of the classifier with these pixel points of the change class and the non-change class as the center.In order to further improve the accuracy of the classifier,the ELM based on genetic algorithm optimization is used to classify,combining the results of the clustering with the results of the GA_ELM,the change detection map of the two-phase SAR image is finally generated.The experiments on multiple real SAR images data show that the proposed method has high robustness to noise,can effectively detect the change of two-phase SAR images with high correct rate and KAPPA value.

关 键 词：合成孔径雷达图像 变化检测 差异图 邻域比 极限学习机 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] TN957[自动化与计算机技术—控制科学与工程]