Nondestructive real-time assessment of sausage quality based on visible-near infrared spectrographic technique

Abstract

Abstract. The visible-near infrared spectrographic technology (vis-NIRS) is greatly used in rapid and nondestructive monitoring of food quality. The sausages is one of most popular meat products, and rapid nondestructive monitoring is an important technical means to ensure its safety. The objective of this paper is to develop a simple and rapid prediction model for sausage corruption. 92 sausage samples were separated into two groups and stored in 4℃ and 20℃, respectively. The visible-near infrared spectroscopy system (350-1100 nm) was used to collect spectral data of each sample from two group each day. Four indicators including total viable bacteria count(TVC), total volatile base nitrogen (TVB-N), color and pH value were used as criteria to access the level of spoilage of the sausages and the values were collected through physical and chemical tests, Two prediction models based on partial least-squares regression (PLSR) and support vector machine (SVM) were established and a comparison was made. Overall, PLS based model with the first derivative preprocessing(FD) of spectral data had better prediction performance. The related coefficient (RC) of TVC , TVB-N , b* and pH indicator were 0.9389 , 0.8849, 0.9097and 0.8899, respectively. To simplified the prediction model, the PLS based model was combine with successive projections algorithm(SPA) which extracting optimum wavelength to improve the detection efficiency. The validation test results revealed that the simplified model could successfully reduce the dimension of the wavelengths and improve detection efficiency while ensured the prediction accuracy in a acceptable rang.