Qualitative Analysis of Lambda-Cyhalothrin on Chinese Cabbage Using Mid-Infrared Spectroscopy Combined with Fuzzy Feature Extraction Algorithms

Abstract

Excess pesticide residues on cabbage are harmful to humans. In this study, we propose an innovative strategy for a quick and nondestructive qualitative test of lambda-cyhalothrin residues on Chinese cabbage. Spectral profiles of Chinese cabbage leaf samples with different concentrations of surface residues of lambda-cyhalothrin were collected with an Agilent Cary 630 FTIR Spectrometer. Standard normal variate (SNV), multiplicative scatter correlation (MSC), and principle component analysis (PCA) were utilized to preprocess the spectra. Then, fuzzy Foley-Sammon transformation (FFST), fuzzy linear discriminant analysis (FLDA), and fuzzy uncorrelated discriminant transformation (FUDT) were employed to extract features from the spectra data. Finally, k-nearest neighbor (kNN) was applied to classify samples according to the concentration of lambda-cyhalothrin residue. The highest identification accuracy rates of FFST, FLDA, and FUDT were 100%, 97.22%, and 100%, respectively. FUDT performed the best considering the combination of accuracy rate and required computing time. We believe that mid-infrared spectroscopy combined with fuzzy uncorrelated discriminant analysis is an effective method to accurately and quickly conduct qualitative analyses of lambda-cyhalothrin residues on Chinese cabbages. This method may have applications in other crops and other pesticide residues.