Novel TVB-N content prediction method of Penaeus vannamei by chromatic value b* and near infrared spectra

Abstract Penaeus vannamei is an important farm-raised shrimp species in China. However, the shrimp deteriorates quickly and has a short shelf life. The surface colour will change with the increase of spoilage, which can reflect its freshness. Therefore, a method for predicting Total Volatile Basic Nitrogen (TVB-N) content using a combination of a chromatic value and Near-Infrared (NIR) spectra is provided. This study explores which chromatic parameter is most efficient for freshness prediction from a data analysis perspective. The Levenberg–Marquardt Optimized Artificial Neural Network (LM-Optimised ANN), the Quadratic Support Vector Machine (Quadratic SVM), and Partial Least Squares Regression (PLSR) algorithms verified the idea. The combination of spectra and chromatic value b* can achieve a more accurate prediction of TVB-N content with RMSEP of 1.660 in Quadratic SVM, 1.337 in PLSR, and of 0.964 in LM-Optimised ANN, better than for other methods. As the polyphenol oxidase blackens the colour of the shrimp, the fusion of NIR spectra and b* can improve the prediction of TVB-N content with fewer parameters. Finally, we constructed an easy and fast TVB-N content prediction system.