Hyperspectral image classification by second generation wavelet based on adaptive band selection

In order to solve problems brought by high dimensions of hyperspectral remote sensing image, a second generation wavelet weighted fusion method based on adaptive band selection (ABS) is proposed in this paper. First, dimensions are reduced by selecting high informative and low correlative bands according to the indexes calculated by ABS method, then, decomposing the selected bands by a novel second generation wavelet, predicting and updating subimages on rectangle and quincunx grids by Neville filters, then using variance weighting as fusion weight, finally the fusion image was classified by maximum likelihood algorithm. AVIRIS hyperspectral data was experimented in order to test the effect of the new method. The results showed classification accuracy is higher after the novel second generation wavelet fusion based on adaptive band selection.