Urban Change Detection of Integrating Remote Sensing and GISTaking Tianjin City for Example

摘要

One Landsat5 TM image of 1993/6/15 and one Landsat7 ETM+ image of 2001/5/12 about Tianjin whose path/row areboth122/33 have been used for this study. An integrated RS and GIS approach is presented for change detection. Basedon summarizing the methods of change detection and analyzing the disadvantages of traditional approaches, multivariatealteration detection based on the canonical correlation analysis is introduced. Firstly, canonical transform is adopted forthe preprocessed images. Then, the sixth component containing the maximal change message is processed and thechange message is extractd. Moreover, the binary image is vectorized and the vectorized maps are overlapped with theoriginal images separately. So the change about two time phases is compared. Subsequently, the database is establishedbased on the basic space data such as road maps and maps showing present condition of land utilization and urbanplanning maps as well as humane and socio-economic data. The results rooting in the image change detection are enteredinto GIS by vectorization and spatial overlay analyzed with already existent data. Finally, the urban built-up area isextracted and the validated precision is high. The urban expansion areas and dynamic change characteristic and reasonsin Tianjin from 1993 to 2001 have been revealed and discussed. Comparing with the Tianjin city master planning(1996-2010), it shows that urban expansion change is coincident with urban planning implementation.