Discrimination of reconstructed milk in raw milk by combining near infrared spectroscopy with biomimetic pattern recognition

Abstract

A new method for discriminating reconstructed milk in raw milk is proposed by combining NIRS (Near Infrared Spectroscopy) with the theory model BPR (Biomimetic Pattern Recognition). In order to compare its discrimination performance, we also carry out experiments by using a traditional method combining near infrared spectroscopy with DA (Discrimination Analysis). The results indicate that the accuracy of detection is 98.3% using the new method; while 87.5% using the traditional method. Comparison results show that the proposed new method is superior to the traditional method and the combination of NIRS with BPR has good potential to detect adulteration of raw milk with reconstructed milk. © 2008 Springer-Verlag Berlin Heidelberg.