Research of Near-Infrared Spectroscopy Method for Wood Density Determination

Abstract

Wood is a remarkably versatile building material. Density is an important indicator of woods properties, which directly impact on the physical, mechanical and processing and using performance. Therefore, it is very significant to research wood density. To predict China fir density, this paper use near-infrared spectroscopy and wavelet transform threshold, wavelet transform modulus maxima and moving variance method to de-noise near-infrared spectroscopy of the first derivative of the wood. We also build wood density partial least squares analysis model and compare them with the model built by the traditional near infrared spectral preprocessing methods. The results show that wavelet transform and moving variance can effectively remove noise in the derivative spectra and improve the accuracy of the spectral analysis model, which means the introduction of Near Infrared Spectroscopy has a great prospect of application in the analysis of building materials.