Hyperspectral image reconstruction method based on regional dynamic depth expansion neural network

Abstract

The invention provides a hyperspectral image reconstruction method based on a regional dynamic depth expansion neural network. The method comprises the following steps: S1, simulating a true value image of hyperspectral data; s2, performing mask coding on the true value image to obtain an aliasing image; s3, performing data preprocessing on the aliasing image, and inputting the aliasing image into a deep expansion neural network for training; the deep expansion neural network comprises a region weight generation module, a threshold iterative algorithm transformation module and a pixel-level adaptive threshold module; and S4, performing spectral image reconstruction by using the trained deep expansion neural network. According to the method, the deep expansion neural network based on regional dynamics is adopted to dynamically guide generation of the reconstruction transform domain according to the regionalization features of the aliasing image, the image reconstruction quality in snapshot compression spectral imaging is effectively improved, the method is more convenient and flexible in network training and practical use, computing resources are saved, and time consumption is reduced.

Description

technical field [0001] The invention relates to the technical field of computational imaging, in particular to a hyperspectral image reconstruction method based on a regional dynamic deep expansion neural network. Background technique [0002] Hyperspectral images are composed of dozens or hundreds of continuous narrow-band images, which can simultaneously capture the spatial and spectral dimensions of the target scene, called "data cubes". With the development of hyperspectral imaging technology, hyperspectral imagers can collect hyperspectral data with higher spatial resolution and spectral resolution. At present, hyperspectral images have been applied and achieved results in many fields, such as ground object remote sensing, precision agriculture, medical diagnosis, target detection, etc. [0003] Snapshot compressed spectral imaging refers to a compressed imaging system that maps multiple frames of spectral images to a measurement value, and its typical system is the Co...

AI Technical Summary

Problems solved by technology

[0005] The present invention aims to provide a hyperspectral image reconstruction method based on regional dynamic deep expansion neural network to solve the problem that the existing snapshot compression spectral image reconstruction method fails to make full use of the characteristic information of the region to guide the reconstruction

Images