Bandwidth function matrix-based spectral deconvolution with alternate minimization method

Spectroscopic data often suffer from spectral distortion and random noise. In this paper, a novel bandwidth function matrix-based spectral deconvolution with the alternate minimization method is proposed to recover the degraded spectra. Different from traditional spectral deconvolution methods which only employ or estimate one bandwidth function, the proposed method introduces a bandwidth function matrix composed of multiple bandwidth functions corresponding to the whole wavelength range of the spectrometer, and applies the alternate minimization method to alternately estimates the real spectrum and bandwidth function matrix until convergence. As a result, the correction error caused by the differences in bandwidth functions in the whole spectral band could be effectively eliminated. A series of experiments was carried out for various simulated and real measured spectra, and comparative results demonstrate the superior performance of the proposed method comparing with the traditional methods.