| CPC G06T 5/60 (2024.01) [G06T 5/50 (2013.01); G06T 7/0002 (2013.01); G06T 2207/10024 (2013.01); G06T 2207/10036 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30168 (2013.01)] | 17 Claims |
|
1. A system for hyperspectral image generation with quality assurance, comprising:
a computing device comprising at least a memory and a processor;
a spectral band grouping module comprising a first plurality of programming instructions that, when operating on the processor, cause the computing device to:
identify a plurality of spectral bands in a training hyperspectral image;
compute a correlation coefficient of each spectral band of the plurality of spectral bands to at least one other spectral band of the plurality of spectral bands; and
form a plurality of spectral domain groups based on the computed correlation coefficients;
a decomposition module comprising a second plurality of programming instructions that, when operating on the processor, cause the computing device to:
obtain the plurality of spectral domain groups from the spectral band grouping module;
obtain an RGB (red-green-blue) input image;
provide the RGB input image and plurality of spectral domain groups to a first neural network, wherein the first neural network includes at least one convolutional block, and at least one residual block; and
obtain as an output of the first neural network, a reconstructed hyperspectral image, based on the RGB input image; and
a quality assurance subsystem comprising a third plurality of programming instructions that, when operating on the processor, cause the computing device to:
obtain the RGB input image, the reconstructed hyperspectral image, and a reconstructed RGB image;
analyze a spectral consistency of the reconstructed hyperspectral image;
evaluate a RGB reconstruction accuracy between the RGB input image and the reconstructed RGB image;
analyze a plurality of noise characteristics in the reconstructed hyperspectral image and the reconstructed RGB image;
generate a plurality of quality scores based on the spectral consistency, the RGB reconstruction accuracy, and the noise characteristics;
compare the plurality of quality scores against a predetermined quality threshold; and
update the first neural network based on the quality score comparisons.
|