Development of Discriminant Analyzing System for Waste Wood Products using Chemometrical NIR Imaging Technique

在本研究中，我们开发了一种新的判别分析方法——近红外化学计量成像，并研究了该技术在木材废弃物中的可行性。采用视像仪采集样品的可见光和近红外图像，并根据校正后的样品与灰板亮度差构建累积直方图。以累积直方图作为伪光谱信息，对三种木材废弃物进行SIMCA分析。SIMCA分析结果显示，所有类别的分类正确率均在80%以上，平均分类正确率为88.9%。SIMCA判别能力分析表明，可见光图像（600 ~ 800 nm）对未处理木材分类有效，近红外图像（1600 ~ 1800 nm）对覆盖木材和未处理木材分类有效。结果表明，该光学系统可作为木材废料回收厂的在线监测系统。

In this study, we developed new discriminant analysis named NIR chemometrical imaging, and examined the feasibility of the technique for wood wastes. Vis and NIR images of sample were captured by vidicon camera, and cumulative histogram was constructed from corrected difference luminance brightness of sample and gray plate. SIMCA analysis was performed for three kinds of wood wastes where cumulative histograms as pseudo-spectral information.The results of SIMCA analysis showed high correct classification of over 80% for all categories, and average correct classification was 88.9%. Discriminating power for SIMCA analysis indicated that visible image (600-800 nm) is effective for classifying non-treated wood, and NIR image (1600-1800 nm) is effective for classifying overlaid and non-treated wood, respectively. These results suggest that the new optical system should be applicable as online monitoring system of recycling mill for wood waste.