Fast information extraction of urban built-up land based on the analysis of spectral signature and normalized difference index
基于谱间特征和归一化指数分析的城市建筑用地信息提取

This paper studies the principles and method of fast remote-sensing information extraction for urban built-up land, taking Fuzhou city as an example. With the detailed analysis and clarity of several existing normalized difference indices, the study selects three indices, Normalized Difference Built-up Index (NDBI), Modified Normalized Difference Water Index (MNDWI) and Soil Adjusted Vegetation Index (SAVI) to represent three major urban land use/cover classes, i. e. , built-up land, water body and vegetation, respectively. The three index images are generated from a Landsat ETM subscene of Fuzhou city and then used as three bands to compose a new image. This dramatically compresses the original eight-band ETM image into a three-band image, reduces band correlations and data redundancy, thus significantly simplifiying the band spectral analysisprocedures. The spectral signature analysis only needs to be performed on this three-index composite image and the signature differences among the three major urban land use/cover classes are revealed much easier than being done with multi-bands. Based on the revealed signature differences, the built-up land is finally extracted through a simple logic calculation. The result achieves a 91.3% overall accuracy. Therefore, the method is a fast and accurate one for the remote-sensing information extraction of urban land use without human interference. In addition to the above built-up land information extraction study, the paper proposes a Modified Normalized Difference Water Index (MNDWI) based on the NDWI of Mcfeeters (1996) , which uses MIR wavelength (ETM band 5) instead of NIR wavelength to construct the index. The replacement largely enhances the contrast of the water bodies with the other land use/cover classes and reduces the spectral confusion with the other classes. Therefore, the MNDWI is more suitable for delineating features of polluted urban rivers/lakes. The advantages of using SAVI instead of NDVI in the urban study are also discussed in this paper.