从TM影像中提取城镇建筑覆盖区专题信息的改进方法

选用TM影像作为城镇建筑覆盖区专题信息提取的数据源，在Bayes的最大似然法半自动提取城镇专题信息和建筑指数法自动提取城镇专题信息分类方法的基础上，通过二值逻辑演算，综合两种方法的分类结果，去掉不属于城镇的信息，得到真正的城镇建筑覆盖区专题信息。实验结果表明，改进后的分类方法，兼有自动与半自动的优点，其分类精度大大提高。特别是对于短期变化(如防洪区)较大、面积较大而复杂的研究区来说，这一方法的应用弥补了单一分类方法的不足。

Modified Method of Extracting Built-up Areas from TM Imagery

This paper uses TM imagery as data sources to extract the built-up areas. Based on Bayes' Maxium Likelihood and Normalized Difference Index method , applying binary calculation, the classification of two methods is synthesized, and those information which doesn't belong to built-up areas is removed, then real thematic information of built-up areas is obtained. Experimental results showed this method makes up a defect of single method. Especially, accuracy of modified classification is largely improved for those regions of short term's change, such as flood control district, and for much bigger and complicate study area.