A New Scheme for Decomposition of Mixed Pixels Based on Modified Nonnegative Matrix Factorization and Genetic-Algorithm

In the decomposition of mixed pixels of hyperspectral remote sensing images, the nonnegative matrix factorization (NMF) easily results in the problem of local minimum, owing to the influence of algorithm initializations. To solve the problem, this paper presents a new scheme based on the modified NMF (MNMF) algorithm and genetic algorithm (GA) to achieve the decomposition of mixed pixels. The endmembers obtained by MNMF is adopted as the initial individual population values of GA, the optimal solution of GA is in reverse as the new initial endmembers in the next running of MNMF, repeat this procedure until the global optimal solution is achieved. Experiment results based on simulated data and real hyperspectral imagery demonstrate that the proposed scheme outperforms NMF and MNMF.