Geodata from Earth HSI bands as a result of direct implementation of Complex wavelet filter Bank for reduction

Abstract

This paper unfolds the distinct feature of complex wavelets to dimension reduction of hyperspectral images. The algorithm of complex wavelet filter bank is implemented on Hyperspectral image bands to reduce the redundant information in terms of bands. In the recent literature on similar concept, the idea is implemented on spectral signatures of each band or in a single layer. The paper encompasses the complex wavelet filter bank advantages related to shift invariance and directionality properties on the hyperspectral image as a whole. The hyperspectral images carry volume of data and the image analysis needs the dimension reduction for the proper computation and detection. The unwanted bands can be ignored through the proposed algorithm. The reduction algorithm realizes the correlation between the bands of 64 band tiff image taken from ISRO database, India. The bands reduced to five in the proposed method efficient than existing real wavelet method.