Space-spectrum fused hyperspectral image mixed pixel low-rank sparse decomposition method

Abstract

The invention discloses a space-spectrum fused hyperspectral image mixed pixel low-rank sparse decomposition method, which comprises the following steps of: normalizing hyperspectral image data, and generating superpixels by adopting a linear iterative clustering method; searching an abundance matrix of a local block of the superpixel, constructing a low-rank constraint expression item, adding a total variation space regular item, and adding a data fidelity item to obtain a target function; introducing a plurality of auxiliary variables to construct a split target function, and converting a constrained optimization problem into an unconstrained optimization problem by adopting augmented Lagrange; and solving the target function by adopting an alternating iterative algorithm to obtain an abundance graph of evaluation confirmation. By adopting the method, the obtained spatial information is more accurate than the simple position relationship among the pixels, the local detail features ofthe abundance matrix are extracted, and the smoothness of the abundance matrix is promoted, so that the whole abundance graph has global features, the target function is split, meanwhile, the algorithm is quickly converged, and a more accurate target solution is obtained.

Description

technical field [0001] The invention relates to the technical field of hyperspectral remote sensing image mixed pixel decomposition, in particular to a low-rank sparse decomposition method of hyperspectral image mixed pixel with spatial spectrum fusion. Background technique [0002] With the development of remote sensing imaging technology, hyperspectral remote sensing images can collect spectral features of hundreds of bands, and have been widely used in agricultural monitoring, military early warning, ground object detection and other fields. However, due to the low spatial resolution and the complex distribution of ground objects, mixed pixels widely appear in hyperspectral images, which largely hinders further exploration of hyperspectral images. The emergence of hybrid pixel decomposition technology greatly overcomes this disadvantage, and lays the foundation for quantitative remote sensing and pixel-level classification. In order to avoid obtaining virtual endmembers ...

AI Technical Summary

Problems solved by technology

However, the existing methods have the following disadvantages: (1) lack of effective spatial information utilization, most methods focus on local windows or first-order neighborhoods as spatial prior information, resulting in low decomposition accuracy; (2) in the abundance The lack of effective correlation constraints on the graph leads to the loss of a lot of detailed information; (3) a lot of regularization parameters are introduced, which makes it difficult to adjust these parameters and affects the final result

Images