Abstract: Raman spectral signals and hyperspectral images of benzoic acid were collected by a Raman hyperspectral imaging spectrometer, and the information was analyzed. The original Raman signal of benzoic acid was preprocessed by a wavelet de-noising method. Optimal parameters for wavelet de-noising that provided the best signal-to-noise ratio (32.092) was determined using an orthogonal array design were established as follows: sym2 wavelet function was used, decomposition level was 2, reset mode was ‘sln’, and threshold option scheme was ‘Rigrsure’. The de-noised Raman spectra were assigned and analyzed. The characteristic vibration modes of benzoic acid in different wavenumber ranges were obtained. The strong spectral peaks at 1 636, 1 603, 1 000, 793, 615 and 420 cm-1 could be used as the Raman characteristic frequency of benzoic acid. The Raman characteristic frequencies corresponding to the gray level image obtained from the hyperspectral image were analyzed. The brightness of the image was correlated with the peak intensity of the characteristic frequency, and both changed in the same order. These research results provide a basis for the detection and analysis of benzoic acid.

Key words: benzoic acid, wavelet de-noising, spectral peak assignment, image analysis