Abstract: This research was based on remotely sensed images of three dates 1992,2000 and 2013 covering Kunming City,Yunnan Province.At first,NDVI image was used to distinguish between vegetation zone and non-vegetation zone.Then,on this basis maximum likelihood classification was applied to each image,getting classified maps of three dates.Based on three classified maps,through analyzing the dynamic changes of land use in 1992-2013,the transition matrix of Kunming City’s land use was developed.Through analysis,the result indicates that the biggest changes of land use in Kunming City were artificial land and unused land,and the growth of artificial land was mainly from unused land.This situation was mainly due to urban extension.

Key words: TM images, land use, dynamic monitoring, transition matrix