多特征差分核支持向量机遥感影像变化检测方法  

Change Detection Method Based on Multi-feature Differencing Kernel SVM for Remote Sensing Imagery

作　　者：林怡[1,2] 刘冰[1,2] 陈映鹰[1,2] 潘琛[3] 

机构地区：[1]同济大学测绘与地理信息学院,上海市四平路1239号20092 [2]同济大学遥感技术应用研究中心,上海市四平路1239号200092 [3]上海市测绘院,上海市武宁路419号200063

出　　处：《武汉大学学报（信息科学版）》2013年第8期978-982,共5页Geomatics and Information Science of Wuhan University

基　　金：国家科技支撑计划后世博科技专项资助项目(2010BAK69B13;2010BAK69B16)

摘　　要：讨论了利用遥感影像光谱、纹理等多种特征信息的多核函数组合方式,给出了多特征空间差分核函数的构建方法,设计了多特征差分核支持向量机变化检测算法,该算法能够实现联合类别样本加权和遥感影像多种变化类别信息的直接检测。实验结果表明,该算法综合利用多种特征信息,检测精度明显高于传统方法,有利于提取小样本的变化信息,避免了以往检测方法需要确定变化阈值的复杂性和不确定性。According to an analysis of support vector machine （SVM） and multiple kernel theory, an improved SVM change detection model based on multi-feature differencing kernel for remote sensing imagery was proposed. The combinations of kernel functions using spec- tral data and textural feature were discussed. After detailing the structure of the image differencing kernel based on multi-features, the algorithm of SVM change detection model was designed, and combined with category weights for the extraction of the spatial distribu- tion of several change classes. Experimental results show that with the help of a multiple kernel function, the change detection model can get higher detection accuracy than the tradi- tional methods, and also avoid determining change thresholds that are complex and uncer- tain.

关 键 词：多特征差分核 变化检测 TM影像 支持矢量机 

分 类 号：TP753[自动化与计算机技术—检测技术与自动化装置] P237.4[自动化与计算机技术—控制科学与工程]