Multilayer non-negative matrix factorization hyperspectral image unmixing method based on spectrum and space total variation minimum limitation

Abstract

The invention discloses a multilayer non-negative matrix factorization hyperspectral image unmixing method based on spectrum and space total variation minimum limitation. The method comprises steps ofsetting a hyperspectral image matrix Y, an end member spectral matrix M, an abundance matrix R and random noise E, and establishing a linear spectral hybrid model to apply multilayer non-negative matrix decomposition to the linear hybrid model; designing a spectral domain and a spatial domain total variation function; introducing the minimum constraint of the total variation of the spectral domain and the spatial domain into MLNMF, and establishing an SSTV-MLNMF target function; optimizing the obtained target function; and selecting experimental parameters to obtain a final unmixing result. According to the method, the characteristic of minimum total variation is applied to a spectral domain and a spatial domain, the effectiveness of the algorithm is verified by simulating hyperspectral image and real hyperspectral image data experiments, and the unmixing precision of the method is higher than that of other methods.

Description

technical field [0001] The invention relates to the technical field of image information processing, and relates to a multi-layer non-negative matrix decomposition hyperspectral image unmixing method with the minimum limit of spectral and spatial total variation. Background technique [0002] Due to the rich band information, hyperspectral images are widely used in satellite remote sensing, crop observation and mineral exploration. However, the spatial resolution of hyperspectral imaging equipment is limited, and the observation distance from the ground is long, so there are mixed pixels in the hyperspectral image, that is, a pixel often contains mixed spectra of multiple types of ground objects. Hyperspectral image unmixing has become an important technical means to distinguish the spectrum (end member) of the object type contained in the mixed pixel and calculate the proportion (abundance) of the object type. Among many unmixing methods, non-negative matrix factorization ...

AI Technical Summary

Problems solved by technology

The traditional non-negative matrix factorization algorithm has only one layer of structure and often cannot adapt to highly mixed hyperspectral images. In recent years, multi-layer non-negative matrix factorization has solved this problem very well.

Images