Using Weighted Filtering to Denoise Low-SNR Spectra Observed through the LAMOST Fiber Optics

In observations with the LAMOST (the Large Sky Area Multi-Object Fiber Spectroscopy Telescope) observed two-dimensional spectral images of low signal-to-noise ratios (SNR) need to be preprocessed to ensure sufficient accuracies of the spectra extracted from the images.In this paper, we propose an improved method for denoising spectral images observed through the LAMOST fiber optics.Our method is based on weighted filtering in the frequency domain of the Wigner transformation.Considering that a conventional low-bandpass filter cannot filter out aliasing noise, we design our method to use the Wigner transformation to obtain spectral profiles of fluxes highly concentrated in the frequency domain.The method first removes appreciable dispersed noise components by a specially constructed bandpass filter, then applies a weighted filter to nonlinearly process the signals in the bandpass based on a priori SNR data.The method finally uses the boundary features of the Wigner transformation to effectively reconstruct signals.We present experiments of applying the improved method to simulated and observed LAMOST data.Our experiments demonstrate the effectiveness of the improved method.