Spectral reconstruction method based on images of RGB color camera

Abstract

The invention provides a spectral reconstruction method based on images of an RGB color camera. The method comprises a training stage and a reconstruction stage. The training stage comprises the following steps: obtaining a first RGB image by the RGB color camera under a light source L1; performing synthesis and white balance transformation computation on the first RGB image to obtain the first RGB image under standard light source LC, and calculating responsivity of the first RGB image; calculating the specific spectral reflection factor of the first RGB image under the standard light sourceLC; performing training and learning on the responsivity and the specific spectral reflection factor of the first RGB image under the standard light source LC to obtain a map. The reconstruction stagecomprises the following steps: obtaining a second RGB image by the RGB color camera under the light source L1; performing synthesis and white balance transformation computation on the second RGB image to obtain the second RGB image under the standard light source LC; calculating responsivity and specific spectral reflection factor of the second RGB image under the standard light source LC by themap to obtain spectral reflectivity of the second RGB image under the standard light source LC.

Description

technical field [0001] The invention relates to the field of spectrum technology, in particular to a spectral reconstruction method based on RGB color camera images. Background technique [0002] Spectral reflectance refers to the ratio of the luminous flux reflected by the object to the luminous flux incident on the object, that is, the relationship between the light reflectance and the wavelength. It is the basis for color measurement, color difference calculation comparison, and computer color matching. In a real lighting environment, the brightness, color, position, and texture of the target scene are often changing, which often leads to changes in the breadth, spectral characteristics, and color appearance attributes. The snapshot spectral imager can obtain the spectral information and spatial information of the instantaneously changing target scene, but the process is complicated and the resolution is low. The spectral reflectance of the surface of an object can also ...

AI Technical Summary

Problems solved by technology

The snapshot spectral imager can obtain the spectral information and spatial information of the instantaneously changing target scene, but the process is complicated and the resolution is low

The spectral reflectance of the surface of an object can also be directly measured by a spectrophotometer and a multispectral camera, but it is difficult for a spectrophotometer to perform non-contact measurement of a non-plane, which limits its practical application

Multispectral cameras, such as multispectral imaging systems that use more than three-color spectral cameras, can obtain accurate spectral reflectance information at each point of an object, but they are expensive and inconvenient to carry, which limits their application in real life

Images