Profile detection for fiber spectrum data with low SNR based on wavelet and filtering

In this paper, a novel detection method is proposed for fiber spectrum profile signals with low signal-to-noise ratio (SNR). This method can be applied to the processing of astronomical fiber spectrum data. First, profile signal of the target is decomposed into several sub-signals by stationary wavelet transform (SWT) algorithm. For sub-signals, adaptive filters are designed with their respective cutoff frequencies by using the corresponding reference signal (flat signal) in order to separate noise from target signal for getting more accurate information of the signal. Inverse stationary wavelet transform (ISWT) is used to reconstruct the detection signal. At the end of this paper, experiments for different full width at half-maximum (FWHM) of point-spread functions (PSF) are presented to demonstrate the effectiveness of the proposed method.