Spectrum estimation method based on adaptive weighted linear regression

Abstract

The technical scheme of the invention is a spectrum estimation method based on adaptive weighted linear regression. According to the invention, a linearized pseudo-inverse solution operator is used asa basis, the digital response value of a camera is expanded by using a homogeneous polynomial, and regularization constraint is carried out on the method by utilizing a Tikhonov regularization method. Firstly, a spectrum estimation method based on a global training mode is established, the homography of the spectrum estimation matrix under different exposure levels can be ensured, and on the above basis, according to the influence mechanism of a training sample on the spectral estimation precision, a Gaussian weighting method is further introduced when a spectral estimation matrix is solved to obtain the spectral estimation method based on the adaptive weighting training mode, so that the adaptive solution of the spectral estimation matrix is ensured, and the spectral estimation precisionis improved.

Description

technical field [0001] The invention belongs to the technical field of computer digital image processing, and in particular relates to a spectral estimation method based on adaptive weighted linear regression. Background technique [0002] The spectral reflectance of the surface of an object (referred to as the spectrum) is used to characterize the ratio of the outgoing light radiation on the surface of the object to the incident light radiation in each band. For objects without special surface properties, the spectrum can reflect the radiation energy of the object itself. The absorption and reflection characteristics of the object are one of the main characteristics of the physical and chemical properties of the object itself. In the range of visible light, the spectrum is the fingerprint of the color information. According to the principle of colorimetry, the relative spectral power distribution, spectrum and color matching function of the light source are analyzed The int...

AI Technical Summary

Problems solved by technology

[0004] For spectral measurements based on digital cameras, calculating high-dimensional spectral data from low-dimensional digital response values ​​is the solution of an ill-conditioned linear inverse problem. Existing spectral algorithms mainly use nonlinear expansion of response values ​​to improve the accuracy of spectral estimation, such as Connah (Connah D, Hardeberg J.2005, Proc Spie 5667), Xiao (Xiao K, Zhu Y, Li C, etal. Optics Express, 2016, 24(13):14934-14950), Zhang (Zhang X, Wang Q, Li J ,etal.Color Research&Application,2017,42(1):68-77) and Liang (Liang J, Wan X.OpticsExpress,2017,25(23):28273-28287) et al.'s research, all used nonlinear polynomial Extended regression method, while the research of Finlayson et al. showed that the spectral estimation method based on nonlinear solution is sensitive to exposure changes, resulting in the lack of generality of the spectral estimation matrix under different exposure levels, which limits the practical application performance of the method

In addition, for spectral measurement based on digital cameras, the closer the spectrum or chromaticity of the training sample and the measurement object is, the more accurate the spectral measurement is. However, most existing methods mainly use the global training method to estimate the spectrum of the measurement object. The accuracy of spectral estimation has been improved, such as the research of Connah and Xiao et al. Zhang et al. proposed a spectral estimation method based on local regression, which improved the accuracy of spectral estimation, but lacked the weighting of samples. Liang et al. proposed a method based on local The spectral estimation method of training sample selection and inverse distance weighting, but training sample selection and weighting are two independent steps, and lack of adaptability to different measurement objects, which still restricts the accuracy of spectral estimation

Images