Abstract
Raman spectroscopy is fast and nondestructive, and it is widely used in chemistry, biomedicine, food safety and other
areas. However, Raman spectroscopy is often hampered by strong fluorescence background, especially in food additives
detection and biomedicine researching. In this paper, one efficient technique was the multi-excitation Raman difference
spectroscopy (MERDS) which incorporated a series of small wavelength-shift wavelengths as excitation sources. A
modified multi-energy constrained iterative deconvolution (MMECID) algorithm was proposed to reconstruct the Raman
Spectroscopy. Computer simulation and experiments both demonstrated that the Raman spectrum can be well
reconstructed from large fluorescence background. The more excitation sources used, the better signal to noise ratio got.
However, many excitation sources were equipped on the Raman spectrometer, which increased the complexity of the
experimental system. Thus, a trade-off should be made between the number of excitation frequencies and experimental
complexity.
