基于资源三号卫星影像的分类器方法研究

Research on the Method of Classifier with ZY-3 Satellite Image

以资源三号遥感影像为数据源,选取老挝首都万象市为试验区,采用面向对象的分类思想,基于不同统计学的分类器算法(决策树算法、支持向量机算法、K最邻近算法、贝叶斯算法)进行地表覆盖要素信息提取,并对不同分类器算法自动解译结果进行精度评价。试验结果表明,采用决策树算法进行自动提取其效果最优,总体精度为93%,Kappa系数为0.91。

The paper mainly use the ZY-3 satellite remote sensing image as data sources,taking the Vientiane city of Laos capital as an example,using the idea of object-oriented classification,based on the different statistics classifier(Support Vector Machine,SVM,K Nearest Neighbor,Bayes)to conduct surface coverage factor information extraction,and evaluating the different classifier algorithm of automatic interpretation result.The experimental results shows that,using the method of support vector machine to conduct the automatic information extraction has the best result.The overall accuracy of information extraction is 93%,Kappa is 0.91.