Soft Independent Modelling of Class Analogy for Near-Infrared Spectroscopic Classifying Some Korean Domestic Conifers

In this study, five species of Korean domestic conifers, which are larch (Larix kaempferi), cedar (Cryptomeria japonica), Korean pine (Pinus koraiensis), red pine (Pinus densiflora) and cypress (Chamaecyparis obtusa) were classified using near-infrared spectroscopy. 50 lumbers for each species were collected from several of the National Forestry Cooperative Federation. Near-infrared absorbance spectra (wavelength range = 680 ~ 2500nm) were acquired on the wide face (radial or tangential face) of each sample located sound parts of heartwood when the lumbers were air-dried (25°C and 65 ± 10% relative humidity). After the acquired near-infrared spectra were mathematically preprocessed (standard normal variate and Savitzky-Golay 2nd derivative), then soft independent modelling class analogy including principal component analysis was performed to classify the species of lumber.