Bayesian-based spectral deconvolution with genetic algorithm

Spectral distortions exist widely due to the broadening effects of spectrometer. In this paper, a spectral deconvolution method using genetic algorithm is proposed to solve this problem. In the Bayesian framework, the spectral deconvolution model is constructed with Gaussian noise and Poisson noise hypothesis, and the prior term is constructed with adaptive Gauss-Markov priori. Genetic algorithm is employed to optimize the spectral deconvolution model to obtain the corrected spectrum. To verify the effectiveness of the proposed method, simulated degraded LED spectra with different colour temperature are corrected by the proposed method, Richardson-Lucy method and Levenberg-Marquardt method. Experimental results show that the proposed method can correct the spectra effectively.