A Study on Selecting Segmentation Object of Object-Oriented Inversion Based on Hyperspectral Chlorophyll-A in Chaohu Lake

Beased on in situ water quality data and hyperspectral data from HJ-1A satellite in Chaohu Lake, through contrasting the object-oriented chlorophyll-a inversion precision of single band with two-band model, the results show: (1) In hyperspectral object-oriented remote sensing inversion, the inversion effect of choosing combination model to segment is superior to that of choosing the single band directly, and using combination model to segment can certain degree solve the problem that commercial softwares cannot segment all hyperspectral data at the same time.(2) When inversing chlorophyll-a concentration with hyperspectral data, the single bands constituting the optimal model are not always in the traditional characteristics band range of chlorophyll-a. So all bands should be comprehensively analyzed to take full advantages of hyperspectral data when inversing. These conclusions will provide basis for the future segmentation object selection of object-oriented on hyperspectral lakes chlorophyll-a inversion and certain reference for band selection of hyperspectral inversion model.