A hyperspectral fusion computational imaging method and system

Abstract

The invention discloses a hyperspectral fusion computing imaging method and system, the invention includes low-resolution hyperspectral image X lr Perform subspace representation to obtain low-rank spectral basis S and low-resolution subspace coefficient A lr ; The low-resolution subspace coefficient A lr After upsampling, it is stacked with the multispectral image Y and input to the pre-trained convolutional neural network to obtain the high-resolution subspace coefficient A; the high-resolution subspace coefficient A and the low-rank spectral basis S are fused to obtain the fused high-resolution hyperspectral Image X. The invention can effectively achieve high-precision quantitative establishment and solution of imaging models of hyperspectral and multispectral images, and generate training data from images to be fused through the established imaging models, solve the problem of insufficient training data, and integrate convolutional neural networks and sub- The combination of spatial representation models reduces network parameters and improves learning efficiency. It has the advantages of lightweight models, high computational efficiency, and high fusion accuracy.

Description

Technical field [0001] The present invention relates to a hyperspectral imaging techniques, particularly directed fusion calculation hyperspectral imaging methods and systems. Background technique [0002] Hyperspectral imaging techniques may acquire a different spectral wavelengths corresponding to the image at the same time, covering the spectral range of the visible band to the short-wave infrared band. Since the reflectivity of different materials is not the same, so the hyper-spectral image can help identify the target accurately, so hyperspectral image is widely used in remote sensing, medical diagnosis, and face recognition and other fields. However, due to the imaging sensor, the imaging optical system existing mutual restraint between the spatial resolution, signal to noise ratio and spectral resolution, high spatial resolution is difficult to directly acquire hyperspectral image, reducing the value of the hyper-spectral image. To improve the spatial resolution of hypers...

AI Technical Summary

Problems solved by technology

This type of method mainly has the following problems: firstly, there is no high-resolution hyperspectral image in practice, which leads to the lack of training data; secondly, in order to achieve higher learning performance, the existing network adopts a deep structure, resulting in low computational efficiency

Images