Abstract: A spectral reflectance estimation method using support vector regression and framclet kernel was proposed.Spectral reflectance estimation is an important subject in optical research. The aim is to convert device-dependent RGB values to device-and illuminant independent reflectance spectra. Regression methods are widely used to estimate spectral reflectance of surface colors given their camera responses, such as regularized least squares method with polynomial models,kernel based regularized least squares method, etc. In this paper, we introduced a novel estimating approach based on the support vector regression method. The proposed approach utilizes a framelet based kernel, which has the ability to approximate functions with multiscale structure and can reduce the influence of noise in data. Experimental resups show that the technicaue can improve the recovery accuracy and stability.

Key words: Support vector regression,Spectral reflectance estimation,Framelet kernel