Hyperspectral Image Denoising base on Adaptive Sparse Representation

Abstract

The hyperspectral image(HSI) generated by most imaging spectrometer systems is constantly disturbed by noise. In this research, we introduced an HSI denoising method based on the theory of sparse coding extended to the spectral domain. The algorithm employs variable separation method and augmented Lagrangian method in sparse coding and constitutes the adaptive dictionary from the pixel spectral vectors extracted from the noisy image. Our results demonstrate that the HSI denoising performance is related to the sparsity of the representation. The effectiveness of the new adaptive sparse coding based approach to hyperspectral denoising, termed HyDeASp, is illustrated in a series of experiments on synthetic and real-world data where it outperforms the state-of-the-art.