Hyperspectral unmixing based on weighted abundance vector regularization

Sparse unmixing is based on the assumption that each mixed pixel in the hyperspectral image can be expressed in the form of linear combination of a number of pure spectral signatures that are known in advance. Despite the success of sparse unmixing based on the L₀ or L₁ regularizer, the limitation of these approach focuses on analyzing the hyperspectral data without incorporating spatial structure information of hyperspectral data. In this paper, considering the correlation between the abundance coefficients of neighboring pixels we proposed a weighted abundance vector sparse unmixing model for the hyperspectral unmixing, named Wav-SU model. Based on this model, we use L₁ norm and L_1/2 norm as the regularizer and adapt the variable splitting and augmented Lagrangian algorithm to solve them. Our experimental results with both simulated and real hyperspectral data sets demonstrate that the proposed Wav-SU method is an effective and simple spectral unmixing algorithm for hyperspectral unmixing.