CAJ | 학술논문

随着遥感技术和成像光谱仪的发展，高光谱遥感图像的分辨率不断提高，其庞大的数据量在提高其遥感探测能力的同时，也给分析和处理带来了很大的困难。高光谱波段选择可以有效减少数据冗余，提高分类识别精度和处理效率。因此如何从多达数百个波段的高光谱图像中选择出具有较好分类识别能力的波段组合是亟待解决的问题。针对上述问题，采用基于图论的谱聚类算法，将原始高光谱图像中的波段作为待聚类的数据点，利用互信息描述两两波段间的相似度，生成相似度矩阵。再根据图谱划分理论，将相似度矩阵生成的非规范化图拉普拉斯矩阵进行谱分解，得到类间相似度小且类内相似度大的类簇；然后根据地物类型计算各波段的类间可分性因子，将其作为类簇内进一步选择代表性波段的参考指标，达到降维的目的；最后通过支持向量机与最小距离分类方法对波段选择后的图像分类。该方法区别于传统的无监督聚类方法，采用基于图论的谱聚类算法，并根据先验知识计算类间可分性因子来选择波段。通过与自适应波段选择算法和基于自动子空间划分的波段指数算法的对比实验，结果表明：两组实验当聚类数目达到相对最佳时，该波段选择方法支持向量机图像总分类精度达到94.08％和94.24％以上，最小距离分类图像总分类精度达到87.98％和89.09％以上，有效保留了光谱信息，提高了分类精度。
수착요감기술화성상광보의적발전，고광보요감도상적분변솔불단제고，기방대적수거량재제고기요감탐측능력적동시，야급분석화처리대래료흔대적곤난。고광보파단선택가이유효감소수거용여，제고분류식별정도화처리효솔。인차여하종다체수백개파단적고광보도상중선택출구유교호분류식별능력적파단조합시극대해결적문제。침대상술문제，채용기우도론적보취류산법，장원시고광보도상중적파단작위대취류적수거점，이용호신식묘술량량파단간적상사도，생성상사도구진。재근거도보화분이론，장상사도구진생성적비규범화도랍보랍사구진진행보분해，득도류간상사도소차류내상사도대적류족；연후근거지물류형계산각파단적류간가분성인자，장기작위류족내진일보선택대표성파단적삼고지표，체도강유적목적；최후통과지지향량궤여최소거리분류방법대파단선택후적도상분류。해방법구별우전통적무감독취류방법，채용기우도론적보취류산법，병근거선험지식계산류간가분성인자래선택파단。통과여자괄응파단선택산법화기우자동자공간화분적파단지수산법적대비실험，결과표명：량조실험당취류수목체도상대최가시，해파단선택방법지지향량궤도상총분류정도체도94.08％화94.24％이상，최소거리분류도상총분류정도체도87.98％화89.09％이상，유효보류료광보신식，제고료분류정도。
With the development of remote sensing technology and imaging spectrometer ,the resolution of hyperspectral remote sensing image has been continually improved ,its vast amount of data not only improves the ability of the remote sensing detec-tion but also brings great difficulties for analyzing and processing at the same time .Band selection of hyperspectral imagery can effectively reduce data redundancy and improve classification accuracy and efficiency .So how to select the optimum band combi-nation from hundreds of bands of hyperspectral images is a key issue .In order to solve these problems ,we use spectral cluste-ring algorithm based on graph theory .Firstly ,taking of the original hyperspectral image bands as data points to be clustered , mutual information between every two bands is calculated to generate the similarity matrix .Then according to the graph partition theory ,spectral decomposition of the non-normalized Laplacian matrix generated by the similarity matrix is used to get the clus-ters ,which the similarity between is small and the similarity within is large .In order to achieve the purpose of dimensionality re-duction ,the inter-class separability factor of feature types on each band is calculated ,which is as the reference index to choose the representative bands in the clusters furthermore .Finally ,the support vector machine and minimum distance classification methods are employed to classify the hyperspectral image after band selection .The method in this paper is different from the tra-ditional unsupervised clustering method ,we employ spectral clustering algorithm based on graph theory and compute the inter-class separability factor based on a priori knowledge to select bands .Comparing with traditional adaptive band selection algo-rithm and band index based on automatically subspace divided algorithm ,the two sets of experiments results show that the over-all accuracy of SVM is about 94.08% and 94.24% and the overall accuracy of MDC is about 87.98% and 89.09% ,when the band selection achieves a relatively optimal number of clusters using the method propoesd in this paper .It effectively remains spectral information and improves the classification accuracy .