[Study on the methods and applications of near-infrared spectroscopy chemical pattern recognition].

Near infrared spectroscopy pattern recognition technique is an important part of modern near infrared spectroscopy technique. In the present paper, main methods of near infrared spectroscopy chemical pattern recognition and some recent developments are introduced. Basic principles of the methods in cluster analysis, discriminant analysis and latent projection are discussed, including some new methods such as support vector machines (SVM), bubble agglomeration algorithm (BA) and focal eigen functions (FEF) etc. Finally, the applications of near infrared spectroscopy pattern recognition technique in agriculture, pharmaceutical industry, food analysis, petroleum industry and other fields are reviewed.