摘要
选取老哈河流域为研究区域,以2007年的两景Landsat5的TM影像为数据源,对该地区进行土地利用/覆盖分类。由于该区域土地覆盖类型复杂,影像较难区分且容易造成错分类。该研究中采用支持向量机(Support Vector Machine,SVM)分类法,通过引入径向基核函数进行非线性变换映射至高维空间,提取它们的非线性特征,增强不同类型之间的可分性,减少错分现象,提高遥感图像分类的精度。通过试验,提取出了2007年的老哈河流域的土地利用/覆盖现状图,以校验该方法的可行性。
The Laoha River catchment is selected as the study catchment.Based on the data source of TM image of Landsat 5 in 2007, classification of the land utilization and covering in the catchment is studied.As the land covering of this catchment is complicated in classification, the images are difficult to separate and easy to classify.In this study, classification method of support vector machine （SVM） is applied.By utilization of radial basis function, the non-linear conversion is conducted to the high-dimensional space, abstrac-ting their non-linear characteristics, strengthening the separation between different types, reducing mistaken classification and improving accuracy of the remote-sense image classification.Through tests, the land utilization and covering status images of the Laoha River catch-ment in 2007 are abstracted to verify the feasibility of this method.
出处
《西北水电》
2015年第3期12-14,共3页
Northwest Hydropower
关键词
老哈河流域
土地利用/覆盖
支持向量机(SVM)
遥感图像分类
Laoha River catchment
land utilization and covering
support vector machine
remote-sense image classification
