摘要
本文以Landsat TM影像数据为基础,采用基于支持向量机分类方法对长白山地区大荒沟林场进行森林植被信息提取,并与传统的最大似然法分类进行对比。结果表明,基于支持向量机方法的森林信息提取精度,Kappa值分别为0.981 0、0.971 6、0.975 3,均超过了最大似然法(MLC)的提取精度和Kappa值0.963 4。该方法有很好的操作性和实用性,准确度满足了林业规划设计的基础数据材料精度要求。
In this paper, the vegetation information of Dahuanggou Forest Farm in Changbal Mountain was extracted by support vector machine （SVM） method based on Landsat TM image. Compared with maximum likehood classification （MLC） whose Kappa value was 0.9634, the classification accuracy and Kappa coefficient （0.9810,0.9716,0.9753） of SVM was higher. The SVM method was useful and practical. Its accuracy satisfies the precision demand of basic data material in forestry programming planning.
出处
《吉林林业科技》
2010年第1期14-17,共4页
Journal of Jilin Forestry Science and Technology
