Rapid authentication of the geographical origin of milk using portable near‐infrared spectrometer and fuzzy uncorrelated discriminant transformation

Abstract

The quality of milk is related to its geographical origin. In addition, the geographical origin of milk plays an important role in its commercial value. Therefore, this research aimed to design an identification system including a portable near‐infrared (NIR) spectrometer and feature extraction algorithms to authenticate the geographical origin of milk. In order to improve the classification accuracy, fuzzy uncorrelated discriminant transformation (FUDT) was used to deal with the NIR spectra of milk collected by a portable NIR spectrometer. In this system, Savitzky–Golay (SG) algorithm and principal component analysis (PCA) were used to remove noise and reduce dimensionality, respectively. Then, three feature extraction algorithms, linear discriminant analysis (LDA), uncorrelated discriminant transformation (UDT), and FUDT were applied to separate the NIR spectra. Finally, K‐nearest neighbor (KNN) classifier was utilized to assess the performance of this identification system. The results showed that the maximum classification accuracies of FUDT, UDT, and LDA were 98.67%, 97.33%, and 93.33%, respectively. The results confirmed the great potential for authenticating the geographical origin of milk by the combination of the portable NIR spectrometer and FUDT. Practical Applications The geographical origin of milk is vitally important to evaluate the milk quality in dairy market. Compared with traditional detection methods, NIR spectroscopy is considered to be a fast, accurate, and nondestructive detection method, so it is widely used in the field of food detection. In this article, FUDT with portable NIR spectrometer can be used to detect the adulteration of geographical origin in milk correctly and rapidly. The experimental result indicated that application potential in identifying the geographical origin of milk.

Rapid Non-Destructive Analysis of Food Nutrient Content Using Swin-Nutrition

“…With the wide application of artificial intelligence technology, deep learning-based food NDDT is a challenging task. Secondly, the current food NDDT technology [ 40 , 41 ] simply predicts and evaluates various food nutrients, and the accuracy needs to be considered. In addition, it is still far from being widely used in the daily life of the public.…”

Rapid analysis technologies with chemometrics for food authenticity field: A review

“…The research on the geographical origin differentiation of animal-derived food products via NIR mainly concentrates on dairy products and seafood. Zhang et al (2022b) distinguished the geographical origins of milk from five Chinese provinces by combining NIR with KNN, showing a discrimination rate of 98.67%. Curro et al (2021) used NIR to analyze the geographical origin of cuttlefish from five FAO (Adriatic Sea, northeastern and eastern central Atlantic Oceans, and eastern Indian and western central Pacific Oceans) fishing regions, with an accuracy reaching 92% after SVM algorithm analysis.…”

Portable NIR spectroscopy: the route to green analytical chemistry

“…Adulteration of goat milk by cow milk was proposed by ( dos Santos et al, 2021 ) and by Muniz and Cuevas-Valdés ( Muñiz et al, 2020 ). To authenticate milk geographical origin, Zhang et al proposed the use of portable NIR spectroscopy to correctly and rapidly detect the adulteration ( Zhang T. et al, 2022 ). To rapidly and conveniently detect the compositions of milk, a handheld detector was developed by Yang et al 120 raw samples helped to set up the PLSR model to predict the main composition ( Yang et al, 2020 ).…”