A Method of Extracting Impervious Surface Based on Rule Algorithm

Shuangyun Peng,Liang Hong,Quanli Xu

DOI: https://doi.org/10.1117/12.2287058

2018-01-01

Abstract:The impervious surface has become an important index to evaluate the urban environmental quality and measure the development level of urbanization. At present, the use of remote sensing technology to extract impervious surface has become the main way. In this paper, a method to extract impervious surface based on rule algorithm is proposed. The main ideas of the method is to use the rule-based algorithm to extract impermeable surface based on the characteristics and the difference which is between the impervious surface and the other three types of objects (water, soil and vegetation) in the seven original bands, NDWI and NDVI. The steps can be divided into three steps: 1) Firstly, the vegetation is extracted according to the principle that the vegetation is higher in the near-infrared band than the other bands; 2) Then, the water is extracted according to the characteristic of the water with the highest NDWI and the lowest NDVI; 3) Finally, the impermeable surface is extracted based on the fact that the impervious surface has a higher NDWI value and the lowest NDVI value than the soil. In order to test the accuracy of the rule algorithm, this paper uses the linear spectral mixed decomposition algorithm, the CART algorithm, the NDII index algorithm for extracting the impervious surface based on six remote sensing image of the Dianchi Lake Basin from 1999 to 2014. Then, the accuracy of the above three methods is compared with the accuracy of the rule algorithm by using the overall classification accuracy method. It is found that the extraction method based on the rule algorithm is obviously higher than the above three methods.