A Nonnegative Matrix Factorization Method Based on Adaptive Local Neighborhood Weighted Constraint and Its Application in Hyperspectral Unmixing

Nonnegative Matrix Factorization (NMF) had become a hot research topic in the field of hyperspectral unmixing. However, it was still difficult to make effective use of hyperspectral space and spectral information, especially when determining local neighborhood, structure fixation was often encountered. To solve the above problems, a non-negative matrix decomposition algorithm based on adaptive local neighborhood weighting constraint was proposed. The local neighborhood of a given pixel could be determined adaptively according to the data characteristics of the abundance. The weight of the algorithm made full use of the spatial and spectral information of the given pixel and the pixel in the neighborhood to improve the performance of hyperspectral unmixing. In this paper, the iterative rule of multiplication was derived by gradient descent method. In order to verify the effectiveness of the proposed algorithm, Japser Ridge data set and Urban data set were used for experiments, and compared with other classical methods, the results showed that this method had better unmixing effect.