基于支持向量机的湿地遥感分类方法

Remote Sensing Classification for Wetland Based on Support Vector Machine

利用遥感的方法监测湿地动态是湿地管理保护的重要手段,本文以陕西合阳黄河湿地为研究区,利用支持向量机的方法提取研究区湿地土地利用现状,并与传统的监督分类方法——最小距离法和最大似然法分类结果进行比较。结果表明,支持向量机方法分类结果总精度高,分类图斑与原始图像一致性高,噪声和错分现象较少,是湿地遥感分类的有效手段。

Remote sensing is an important way to realize wetland dynamic monitoring, and play an effective role of management and protection of the wetland. With Yellow River wetland in Heyang of Shaanxi province as the study area, support vector machines （SVM） method is applied to retrieve land use/land cover types, and compare the result with traditional supervised classification methods （Minimum Distance and Maximum Likelihood）. The result shows that SVM classification method has higher classification accuracy and lower misclassification rate, the consistency of the classification result and original image are high, and it is efficient tools in wetlands classification.