Fast Detection Method of Antarctic Krill Meat Quality Based on Near Infrared Spectroscopy

  • Lanlan Zhu
Keywords: Near-infrared Spectroscopy, Antarctic Krill, Rapid Detection, Data Processing, Qualitative Analysis Model

Abstract

In order to protect the health, safety and legal rights of consumers, the quality of shrimp meat must be
strictly controlled during the processing, transportation and marketing of shrimp meat. Traditional shrimp meat
quality testing is mainly based on chemical methods, which often require multiple chemical instruments and
reagents. Samples require pretreatment, which is cumbersome and time-consuming. Based on the above
background, the purpose of this paper is to study the rapid detection method of Antarctic krill meat quality based
on near-infrared spectroscopy analysis technology. In this paper, a qualitative analysis model of shrimp meat
quality using near-infrared spectroscopy is established based on four methods: support vector machine, BP
neural network, random forest and width learning. Combined with the principles of modeling methods, the
performance of the analysis model is comprehensively evaluated under the same preprocessing method and
sample set partition conditions. The best prediction accuracy of the five-fold cross method of the four qualitative
analysis models reached 88.48%, 88.57%, 89.05%, and 79.62%, respectively. Then, based on the existing
theoretical methods, a method for rapid detection of shrimp freshness using a portable near-infrared
spectrometer is studied. By comprehensively assessing the two indicators of sample accuracy and time, a
combination mode based on the standard normal variable transformation + discrete Fourier transform method +
support vector machine method is determined to construct a qualitative analysis model of the near red spectrum
of shrimp meat to achieve real-time online discrimination fresh and refrigerated shrimp. In order to accelerate
the process of near-infrared spectroscopy used to achieve fast online detection of shrimp quality.

Published
2020-03-01