Near-IR material discrimination method by using multidimensional response variables PLS regression analysis

Abstract

In this paper, we propose a method to discriminate human skin, plants, and asphalt based on the spectral reflectance properties of each material in near-infrared ray (NIR) band. A single material is detected by using only two or three wavelengths corresponding to the absorption bands and the other wavelength based on the spectral reflectance properties of the target material. However, with more than two materials, materials are not always discriminated by using only the absorption wavelengths of each material. In this paper, we discriminated three materials using multiple NIR wavelengths. In this proposed method, a multiclass classifier for the materials is generated by using Partial Least Squares (PLS) regression analysis, and three materials are discriminated by multiple predicted values obtained by the classifier. Also, optimum wavelengths to discriminate are selected by using VIP. As a result of an experiment, three materials were discriminated with high accuracy.