基于SPOT5影像针叶林信息提取——以攸县黄丰桥林场为例

Coniferous Forest information Extraction Based on the SPOT5 Images——Case of Huangfengqiao Forestry Farm

为了探究基于SPOT5影像的针叶林信息提取的方法,提高森林资源调查效率,为森林资源的合理开发和科学规划提供依据,以湖南省攸县黄丰桥国有林场为研究对象,利用SPOT5遥感影像及外业调查数据为数据源,运用最大似然分类、神经网络分类法、马氏距离分类法3种分类法对针叶林信息进行提取,并对其精度进行评价。研究结果表明:总体精度最高的是最大似然分类法89.223 0%,其次是神经网络分类法86.068 2%,马氏距离分类法是79.737 5%;Kappa系数分别为0.865 3,0.825 5,0.746;最大似然法在保证针叶林分类精度时,其它类型森林分类精度也比较高,能达到较为理想的效果。

In order to explore the way of the coniferous forest information extraction, and improve the efficiency of forest resources survey, and provide a basis for the rational development of forest resources and scientific planning, Took Huangfengqiao forestry, farm of you county as the research object, the paper used the SPOT5 re mote sensing images and field investigation data as the data source, the coniferous forest information was extrac ted by the maximum likelihood classification, neural network classification, Mahalanobis distance classifica tion, to evaluate its accuracy. The results showed that： tile maximum likelihood classification, minimum dis tance classifier, Mahalanobis distance classification overall accuracy were 89.223 0% , 86.068 2%, 79. 737 5%. Kappa coefficient were 0. 865 3,0. 825 5,0. 746 5. Maximum likelihood method could guaranteed the coniferous forest classification accuracy, meanwhile other types of forest classification accuracy were rela tively high, they could also be effective methods.