Related Experiment Video
Updated: Mar 6, 2026

O-cresol Concentration Online Measurement Based On Near Infrared Spectroscopy Via Partial Least Square Regression
Published on: November 8, 2019
Comparative analysis of discrimination accuracy across different near-infrared bands for rosewood species
Yuanda Qi1, Guangqiang Xie1, Yaoxiang Li1
1College of Mechanical and Electrical Engineering, Northeast Forestry University, Harbin 150040, PR China.
Abstract:
Rosewood has long been prized worldwide for its warm color, distinctive grain, high density, and excellent mechanical properties. It has a long history of use across many cultures and regions. With global rosewood resources in decline, counterfeiting, adulteration, and smuggling have become routine in the rosewood trade, underscoring the need for a rapid, non-destructive, and accurate means of species authentication. Near-infrared spectroscopy shows significant promise in wood identification, yet there is still no consensus on the optimum spectral window. To address this gap, we compared four commonly applied wavelength intervals-visible-near-infrared (Vis-NIR, 350-2500 nm), short-wave NIR (SW-NIR, 700-1100 nm), truncated long-wave NIR (tLW-NIR, 1100-2500 nm) with the noisy tail from 2150 nm to 2500 nm was removed, and a truncated full-range NIR (tFULL-NIR, 700-2150 nm) with the noisy tail from 2150 nm to 2500 nm was removed in its entirety. Eight commonly used rosewood species listed in the Chinese National Standard were analyzed: Pterocarpus santalinus, Dalbergia bariensis, Pterocarpus erinaceus, Pterocarpus macrocarpus, Millettia laurentii, Dalbergia cochinchinensis, Dalbergia louvelii, and Dalbergia retusa. After seven preprocessing routines had been applied, every "window + preprocessing" combination was evaluated with partial least-squares discriminant analysis (PLS-DA), support-vector machines (SVM), and long short-term-memory (LSTM) networks. The tLW-NIR delivered the highest overall accuracy (98.53%) when smoothed with Savitzky-Golay (S-G) filtering followed by Mean centering (S-G + Center) and classified by LSTM. NIR with band optimization coupled with LSTM is a useful strategy for rapid and non-destructive discrimination of rosewood. The study offers a practical guidance on wavelength-window selection and model generalization, and provides a robust technical tool for combating illicit activity in the rosewood trade.
More Related Videos
Related Concept Videos
IR Frequency Region: Fingerprint Region
Attenuated Total Reflectance (ATR) Infrared Spectroscopy: Overview
The ATR process begins by directing a beam...

