A study of Fourier denoising and wavelet denoising in the reconstruction of cellular and serum Raman spectra

Raman spectroscopy is widely used in biomedicine, material science and other fields, but its signal quality is often affected due to the interference of background fluorescence and instrumental noise, and the choice of denoising method is crucial to the accuracy of spectral reconstruction. In this paper, the effects of two different denoising methods, Fourier denoising and wavelet denoising, in spectral reconstruction are systematically investigated. The experiments are based on both cellular and serum samples, using weighted Wiener estimation for reconstruction, and combining different denoising methods to optimise the signal quality, and finally using the root-mean-square error as the evaluation criterion. The results show that in the reconstruction of the two materials, Fourier denoising can improve the reconstruction accuracy by 3% and 21.9%, while wavelet denoising can improve the reconstruction accuracy by 0.9% and 28.2%, respectively. This study provides an important reference for the optimisation of spectral signals and helps to improve the accuracy of Raman spectroscopy for applications in complex environments.