论文部分内容阅读
用模式识别技术进行森林分类,可大面积、快速地取得有关森林分布的信息,分类精度可达80%以上。本试验是在牡丹江地区江山娇林场和带岭林业局凉水林场进行的,采用美国第二号地球资源技术卫星1981年7月13日和1975年9月2日提供的卫星影象资料。一、试验方案的选取方案的选取以提高分类精度、加快分类速度为依据。采用无监督分类和有监督分类两种方法相结合。首先用无监督方法处理一部分象元(由于数据量小,小机器也可胜
Forest classification using pattern recognition technology can obtain information about the distribution of forests in a large area and rapidly, and the classification accuracy can reach more than 80%. This experiment was carried out in Jiao Linchang, Jiangshan and Liangshui Forest Farm of Zhaoling Forestry Bureau in Mudanjiang. The satellite image data provided by the USSII satellite on July 13, 1981 and September 2, 1975 were used. First, the selection of the test program selection program to improve the classification accuracy, speed up the classification based on speed. The combination of unsupervised classification and supervised classification. First, using unsupervised methods to deal with part of the pixel (due to the small amount of data, small machines can win