Abstract

Hyperspectral image (HSI) processing is one of the key processes in satellite imaging applications. Hyperspectral imaging spectrometers collect huge volumes of data since the image is captured across different wavelength bands in the electromagnetic spectrum. As a result, compression of hyperspectral images is one of the active area in research community from many years. The research work proposes a new compressive sensing based approach for the compression of hyperspectral images called SHSIR (Sparsification of hyperspectral image and reconstruction). The algorithm computes the coefficients of fractional abundance map in matrix setup, which is used to reconstruct the hyperspectral image. To optimize the problem with non-smooth term existence along with large dimensionality, Bregman iterations method of multipliers is used, which converts the difficult optimization problem into simpler cyclic sequence problem. Experimental result demonstrates the supremacy of the proposed method over other existing techniques.