Extracting of remote sensing information of wetland in Hangzhou Bay Area in Zhejiang Province based on multi-layer classifying technique

Taking Hangzhou Bay Area in Zhejiang Province as the study area and using Landsat TM images as the data source, this paper brought forward the multi-layer classifying technique on the basis of spectral and spatial characters analysis of all types of features. Firstly, the Modified Normalized Difference Water Index (MNDWI) modal was used to extract the water bodies. Then, the unsupervised classification technique and the interactive interpretation technique according to the shape character and spatial distribution of the features were used to extract all other types. The results showed that the total precisions of the three periods of images were all above 90%.