Study on the key technology of spectral reflectance reconstruction based on the weighted measurement matrix

Abstract

In order to reconstruct the spectral reflectance accurately, a new method of spectral reflectance reconstruction based on the weighted measurement matrix is proposed in this paper. By optimizing the measurement matrix between spectral reflectance and the response of a camera, the method can improve the reconstruction accuracy. The new method is a combination of three kinds of common reflectance reconstruction methods, which are the pseudo inverse method, the Wiener estimation method and the principal component analysis method. The new measurement matrix can be achieved after weighting the measurement matrices of these three methods to reconstruct the spectral reflectance. What is more, the weights of the three methods can be obtained by minimizing the color difference. Results show that the CIE1976 color difference and RMSE value of the weighted reconstructed spectra are less than that of three common reconstruction methods. The spectral matching accuracy GFC of the method is higher than 0.99 and its reconstruction accuracy is high.