在分析传统统计模式识别分类方法分类精度不高的现状的基础上,以OMIS—I影像为例,采用基于支持向量机的方法对延河流域枣园地区植被信息进行提取,取得了很好的实验结果。与传统的最大似然分类提取方法相比,基于支持向量机的方法提取精度达90.50%,Kappa系数也超过了0.87,比单纯的最大似然分类方法提取精度高得多,而且该方法具有很强的操作性和实用性。图6,表2,参6。
On the basis of analyzing the actuality of the low accuracy in the traditional statistical pattern recognition classification, the principle and application of support vector machine (SVM) is introduced with a real case of OMIS-I data. Using the OMIS-I data of ZaoYuan , an experiment is conducted and an excellent result is gained. Compared with the traditional Maximum Likelihood Classification (MLC) method, the resuh shows that the precision of this method reaches 90.50 %, kappa coefficient exceeds 0.87. Thus, this method has more superiority and practicability in Hyperspectral remote sensing image classification.