Abstract: Spectroscopic data of wood samples gathering by spectroscopic instruments are disturbed by a series of noises and interferences,therefore the proper data preprocessing is very important for model establishment and achievement of accurate analytical result. This paper was direct application of wavelet transform combined with first derivative in spectrum preprocessing of Chinese Fir. Using the db3 wavelet to resolve spectrum data in four races. In addition,according to the different characteristics of transformation between signal and noise in different resolve races,to hold back the maximum of signal model and dislodge the minimum of the noise model.The results showed that the combination of first derivative and wavelet transform modulus maximum was able to eliminate spectroscopic noises and interferences as well as reserve major information. It contributed to increase analysis quality and precision of the near infrared.

Key words: near infrared spectroscopy, wavelet transform, modulus maximum, denoising