A Spectral Reconstruction Method Based on RGB Values

Abstract

The invention discloses a spectral reconstruction method based on RGB values, comprising the following steps: (1), preparing a training sample set, obtaining its RGB values ​​and spectral reflectance values, constructing the correlation between the two according to the spectral projection formula, and obtaining a projection matrix ; (2), obtain the spectral dictionary matrix X; (3), prepare the test sample set, obtain its RGB value and construct the relational expression between it and the spectral dictionary matrix X, and solve the spectral coefficient b in the relational expression in conjunction with the A projection matrix; ( 4), Substituting the spectral coefficient b and the spectral dictionary matrix X into the spectral reconstruction formula to complete the spectral reconstruction. The present invention obtains the spectral dictionary matrix X, so that the reflectance spectrum of any color can be linearly represented by the spectral coefficient b and the spectral dictionary matrix X, so as to reduce the possibility of pathological conditions.

Description

technical field [0001] The invention belongs to the technical field of spectral reconstruction, and in particular relates to a spectral reconstruction method based on RGB values. Background technique [0002] Compared with traditional color images, spectral images can obtain useful information with richer content and more prominent details, and spectral imaging technology has developed rapidly. In recent years, some scholars have carried out research on spectral reconstruction methods based on RGB digital cameras. The spectral imaging system makes it possible to obtain spectral images with simple, portable, and low-cost ordinary cameras. However, for this spectral imaging system, reconstructing high-dimensional spectral reflectance information from low-dimensional digital response signals (RGB values) is a mathematical solution process of an ill-conditioned linear system. The coefficient matrix of this linear system is a low-rank matrix and The number of rows and columns o...

AI Technical Summary

Problems solved by technology

However, for this spectral imaging system, reconstructing high-dimensional spectral reflectance information from low-dimensional digital response signals (RGB values) is a mathematical solution process of an ill-conditioned linear system. The coefficient matrix of this linear system is a low-rank matrix and The number of rows and columns of the matrix varies greatly, resulting in huge problems in reconstruction accuracy and stability when using this linear system to reconstruct multidimensional spectra

Images