用陆地卫星TM数据对农作物进行自动分类的研究

文中阐述了采用1985年7月4日和8月5日的陆地卫星TM数据对纽约州西里卡县进行农作物自动分类的研究。农作物主要指玉米和小麦。使用监督的最大似然率数字图象分类法。分类结果的精度:玉米为72～91％的正确分类率,小麦为82～88％的正确分类率,同时分类的附加误差很小。文中对如何选择训练数据,以提高分类的精度和可靠性作了研究。

Analysis of Thematic Mapper Satellite Data for Automatic Classification of Field Crops

An analysis for automatic classification of field crops-field corn and wheat, using the Landsat Thematic Mapper (TM) scenes on July 4 and August 5, 1985, in Seneca County, New York state is presented. Emphasis is placed on digital image classification using a maximum likelihood classifier. The results of this TM data classification show that the ranges of accuracy of digital image with a maxi- mum likelihood classifier are from 72% to 91% for corn and from 82% to 86% for wheat. Classification of field crops has low errors of commission. The paper also reports how to choose the training data to obtain more accurate, reliable classification.