论文部分内容阅读
遥感技术在大尺度土壤盐渍化检测方面有着宏观性、实时性、动态性等优势和广阔的应用前景,但是传统的遥感图像分类方法精度不高、分类效率较低和不确定性。提出了基于支持向量机(Support Vector Machine,SVM)的分类方法,介绍了SVM算法的基本原理,通过支持向量机分类法与传统分类方法(最大似然法和最小距离法)在盐渍化信息提取结果上进行对比,表明基于SVM的遥感图像分类方法能够较好的检测土壤的盐渍化信息,分类总精度达到95.66%,比最大似然法和最小距离法分类精度(分别为91.54%和85.42%)更高,因此更适合于遥感图像分类和盐渍化信息检测。
Remote sensing technology has the advantages of macroscopical, real-time, dynamic and other applications in large-scale soil salinization detection. However, the traditional remote sensing image classification method is not accurate, the classification efficiency is low and uncertain. A classification method based on Support Vector Machine (SVM) is proposed. The basic principle of SVM is introduced. By using support vector machine (SVM) classification and traditional classification methods (maximum likelihood method and minimum distance method) The results show that the remote sensing image classification method based on SVM can detect the salinization information of soil well, the total accuracy of classification is up to 95.66%, which is 91.54% and the classification accuracy of maximum distance method and minimum distance method respectively 85.42%), so it is more suitable for remote sensing image classification and salinization information detection.