Abstract: The kernel function is key part of Kernel Principal Component Analysis, KPCA. The present used kernel functions are simple kernel functions. This paper makes an effort to present a mixed kernel function by combining Spectral Angle Radial Basis Function, SA-RBF with RBF. In this study, extracting spectral feature bands using KPCA based on the mixed kernel function, the SVM is used to classify salt-affected soil using a combination of spectral features and texture features as a data source. In addition, the combined approach is compared with other SVM methods. The results reveal that the proposed SVM method used here can effectively extract salt-affected soil thematic information for the Manasi River Oasis. Especially, the overall accuracy of this method is 89.000% and the kappa coefficient is 0.876, which indicates that this method is better than other classification methods.

Key words: Mixed Kernel Principal Component Analysis, texture features analysis, Support Vector Machine（SVM）, salt-affected soil

摘要： 核函数是核主成分分析（Kernel Principal Component Analysis，KPCA）的核心，目前使用的核函数都是单一核函数。尝试通过将光谱角径向基核函数（Spectral Angle Radial Basis Function，SA-RBF）与RBF组合形成混合核函数。在研究中，利用基于该混合核函数的KPCA进行特征提取，将其光谱特征波段和纹理特征相结合用于盐碱土的SVM分类，将分类结果与其他SVM分类进行比较，结果表明：该方法优于其他SVM方法，能有效提取玛纳斯河流域绿洲区的盐碱土专题信息，分类精度是89.000%，kappa系数是0.876。

关键词: 混合核主成分分析, 纹理特征分析, 支持向量机, 盐碱土