Spectral reflectivity reconstruction method combining principal component analysis and regularized polynomial

Abstract

The invention discloses a spectral reflectivity reconstruction method combining principal component analysis and a regularized polynomial. The method comprises the following steps: 1) collecting spectral data of training samples by employing a multi-spectral imaging system; 2) carrying out dimensionality reduction on the spectral data of the training samples by means of a principal component analysis method to reduce the spectral data quantity of the training sample; and 3) constructing a training sample set through the spectral data of the training samples processed in the step 2), then carrying out polynomial regression channel responding expansion on the training sample set, then constructing a target function by taking the minimum error between the reconstructed spectral reflectivity and the actual value as a target and adding constraining items into the target function by employing a Tikhonov regularization method, and finally, solving the target function to obtain the reconstructed spectral reflectivity so as to reconstruct the spectral reflectivity combining the combining principal component analysis and the regularized polynomial. The method can be used for reconstructing the accurate spectral reflectivity.

Description

technical field [0001] The invention belongs to the field of digital image processing, and relates to a spectral reflectance reconstruction method combining principal component analysis and regularization polynomial. Background technique [0002] The traditional method of obtaining the spectral reflectance of an object is to use a spectrophotometer to measure point-to-point. The multispectral imaging system can be used to collect the multispectral color information of the object under multiple channels, and then the reconstruction algorithm can be used to efficiently reproduce the continuous spectrum of the object surface. This method is called the spectral reflectance reconstruction method based on multispectral imaging technology. This method can simultaneously obtain the information of the target object from the spectral dimension and the spatial dimension. However, the dimension of the traditional chromaticity space is only one-tenth of the dimension of the spectral spa...

AI Technical Summary

Problems solved by technology

However, the dimension of the traditional chromaticity space is only one-tenth of the dimension of the spectral space. When using the spectral reflectance to represent the color, a large amount of spectral data needs to be processed, which makes the processing efficiency low and the storage space increases. Therefore, when the reconstruction accuracy is met Under the conditions, data compression can be performed on a large amount of spectral data to achieve the effect of spectral dimensionality reduction

At present, commonly used spectral data dimensioning methods include principal component analysis, independent principal component analysis and other improved correlation algorithms, but none of the above methods can reconstruct accurate spectral reflectance

Images