Diagnosis of food safety based on pattern discrimination of attenuated total reflectance IR spectrum

This paper describes a sophisticated diagnosis method of pesticide residues in the agricultural products. The proposed method uses the attenuated total reflectance of infrared light on the sample surface. Lettuces were obtained in the real field as the sample set and were categorised into three groups by the cultivation procedures of tetrachloroisophtharonitrile fungicide application. The categorisation for each sample was verified by chromatography and the results were used as the reference values of residues. Attenuated total reflectance spectrum of each sample lettuce was measured by FTIR spectrometer equipped with a contact sampler. The K-nearest neighbour method and the soft independent modelling of class analogy method have derived two kinds of diagnosis models. These models were verified by using 31 ATR spectra of the validation sample