Ces: a new stellar spectral noise reduction algorithm

We propose a stellar spectral noise reduction algorithm named Ces. The Hermitian interpolation algorithm is utilized to replace the cubic spline interpolation module in the empirical mode decomposition algorithm. Additionally, the shape of both upper and lower envelopes are constrained by continuous spectrum information from celestial spectra. The specific steps involved are: (i) The continuous spectrum of stellar spectra is fitted using a 10th-order polynomial. (ii) Calculate the derivative of the continuous spectrum at the extreme point of the stellar spectrum. (iii) Input the calculated deri­vative into the Hermite interpolation algorithm, utilizing it as a fitting parameter to generate upper and lower envelopes. Ces algorithm firstly decomposes and reconstructs stellar spectra to achieve initial noise reduction effect. The singular value decomposition algorithm is used to process the initial noise reduction data again to further remove the noise and recover part of the spectral line information.