Hyperspectral image reconstruction method based on double-ghost attention mechanism network

Abstract

The invention provides a hyperspectral image reconstruction method based on a double-ghost attention mechanism network, which is used for performing hyperspectral image reconstruction on a single RGB image, and comprises the following steps: step 1, inputting and processing an RGB image to obtain shallow feature information; 2, enabling the shallow feature information to enter a first double-ghost residual attention module, and obtaining deep feature information; 3, enabling the deep-level feature information to enter a dual-output feature convolution attention mechanism module to extract deeper-level feature information; 4, adding the deeper feature information and the shallow feature information to obtain new features; 5, putting the new features sequentially into subsequent m-1 series double ghost residual attention modules, and finally outputing one feature; 6, performing convolution on the features output in the step 5, and activating to obtain new features; and 7, enabling the feature map with the new features to enter an optimal non-local module, and outputting a hyperspectral image for visualization.

Description

technical field [0001] The invention relates to an image reproduction method, in particular to a hyperspectral image reconstruction method based on a double-ghost attention mechanism network. Background technique [0002] Hyperspectral imaging technology is based on a large number of narrow-band image data technologies. It combines imaging technology with spectral technology to detect the two-dimensional geometric space and spectral information of the target, and obtain high-resolution continuous, narrow-band image data. Hyperspectral images combine image information and spectral information of samples. Image information can reflect the external quality characteristics of the sample such as size, shape, defect, etc. Since different components have different spectral absorption, the image will reflect a certain defect more significantly at a specific wavelength, and the spectral information can fully reflect the quality of the sample. The difference in internal physical stru...

AI Technical Summary

Problems solved by technology

However, the accuracy of the reconstructed hyperspectral image needs to be further improved. At the same time, the trained model is computationally complex and takes up a lot of memory. This is also a common problem in the current RGB reconstruction of HIS images.

Images