摘要
采用2007年6月云南省勐腊县TM遥感数据,利用EnMAP-box进行了支持向量机的图像分类研究,以网格搜索法寻找最优参数,在设定的范围内,求得了最优C和g参数,用此参数进行支持向量机的遥感图像土地覆盖分类。结果表明:SVM方法较最大似然分类方法具有较高的分类精度,特别是阔叶林和橡胶林的精度明显优于最大似然分类方法;对于面积较小的次要类型,2种分类方法的精度基本保持一致;SVM的总体精度相对于最大似然分类提高了11.9%。
Image classification of the TM remote sensing data of Mengla County,Yunnan Province in June of 2007 was conducted by EnMAP-box model with the support vector machine （SVM）,attempting to search for the optimal parameters by grid search.The optimal C and g parameters were obtained within a set range,and the land cover classification was done by SVM with the optimized parameters and the remote sensing image.The results showed that the classification accuracy of SVM classifier was higher than that of the regular Maximum Likelihood Classifier （MLC）,especially for the broadleaved forests and rubber plantations.The classification accuracy of the two methods would be similar for smaller secondary land types.Comparatively speaking,the overall accuracy of the SVM was 1 1 .9% higher than that of MLC.
出处
《西南林业大学学报（自然科学）》
CAS
2014年第2期67-71,共5页
Journal of Southwest Forestry University：Natural Sciences
基金
国家公益性行业科研专项(200904045)资助
国家自然基金项目(31260156)资助
关键词
支持向量机
EnMAP—box
网格搜索法
遥感图像分类
support vector machine （SVM）
EnMAP-box Model
grid search
remote sensing image classification
