Spectral Unmixing Using a Sparse Multiple-Endmember Spectral Mixture Model

In remote sensing data exploitation, the spectral mixture analysis technique is generally used to detect the land cover materials and their corresponding proportions present in the observed scene. Traditionally, a fixed endmember spectral signature for each land cover material is used to perform the unmixing task. In the literature, some scholars have proposed performing the unmixing by taking the spectral variability into consideration. Among these spectral-variability-based unmixing approaches, multiple-endmember spectral mixture analysis (MESMA) is probably the most widely used method. However, when the number of land cover materials is large, the computational load of the MESMA method could be very heavy. In this paper, a sparse multiple-endmember spectral mixture model (SMESMM) is proposed to handle this problem. This model treats the spectral mixture procedure as a linear block sparse inverse problem. The SMESMM is first solved using a block sparse algorithm to obtain an initial block sparse solution. Then, MESMA is used to resolve the mixed pixel using the selected land cover materials, which correspond to the nonzero blocks in the solution obtained in the first step. The block sparse solution obtained in the first step can help to determine how many and which land cover materials are involved in the considered mixed pixel. This can largely decrease the number of possible candidate models for the MESMA method when the number of land cover materials is large. Experimental results on simulated and real hyperspectral data demonstrate the efficacy of the proposed method.