Reconstruction method of snapshot spectral imaging system based on tensor low-rank constraint

Abstract

The invention discloses a reconstruction method of a snapshot spectral imaging system based on tensor low-rank constraint, and belongs to the field of computational photography. The method is appliedto two snapshot spectral imaging systems, namely, a coded aperture snapshot spectral imaging system and a dual-camera spectral system based on a panchromatic camera. The reconstruction method comprises the steps of firstly, constructing a three-dimensional tensor by using the non-local similarity of a hyperspectral image, then mining structural characteristics of the three-dimensional tensor by using a dimension-discriminative low-rank tensor regularization (DLTR) model, wherein the structural characteristics include the spatial self-similarity, the spectral correlation and the space-spectrumjoint correlation; and finally, alternately updating and iteratively solving so as to complete the high-precision hyperspectral image reconstruction. According to the invention, the high-dimensional physical characteristics of the hyperspectral image can be better transmitted, the internal structural characteristics of the hyperspectral image can be better mined, the reconstruction quality of thesnapshot spectral imaging system is greatly improved, and the method has the advantage of high reconstruction precision.

Description

technical field [0001] The patent of the present invention relates to a hyperspectral image reconstruction method for a snapshot spectral imaging system, especially a method capable of reconstructing a hyperspectral image with high precision, which belongs to the field of computational photography. Background technique [0002] Hyperspectral imagery is capable of acquiring continuous spectral sampling of the target scene. Compared with RGB color images or panchromatic images, hyperspectral images can provide rich spectral information and detailed structures. At present, hyperspectral images have been applied to many fields such as environmental monitoring, geological exploration, aerospace and medical detection. [0003] A hyperspectral image is called a 3D data cube, which contains two-dimensional spatial information and one-dimensional spectral information. In order to obtain hyperspectral images, traditional imaging spectroscopy systems use scanning methods to obtain sp...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the problem that existing reconstruction algorithms ignore the high-dimensional

Images