Non-destructive Sensing Of Quality Changes In Beef During Microwave Heating Using Hyperspectral Imaging

Beef contains rich nutrients.The use of microwave to treat beef has increased in industrial production and domestic household,because of its advantages of high efficiency and better nutrient retention in food products.However,it is challenge to monitor microwave heating process.Thus,it becomes significant to improve the quality attributes and monitor parameter changes of beef during microwave heating process.This study was designed to track the changes of various quality attributes of beef during microwave heating process,including moisture content,color,tenderness,and proteins using a visible/near-infrared(400-1000 nm)hyperspectral imaging system.Besides,the conception of doneness degree of beef during microwave heating was proposed,and the HSI technique successfully achieved classification of doneness degree as well.The encouraging results demonstrated that the ability of HSI system for monitoring the changes of some quality parameters of beef during microwave heating.Specifically,the main research results are shown as follows:(1)Microwave heating time significantly influenced MC and color of beef,and the HSI was succussfully used to predict MC and color.The results showed a significant increase of L*value and decrease of MC and a*value during the later stage of microwave heating(45-70 s).For predicting MC,the SG-SPA-LS-SVM model achieve best result with R_P~2 of 0.869 and RMSEP of 1.304.For predicting a*,the SG-RC-MLR model showed best result with R_P~2 of0.890 and RMSEP of 0.735.The models were used to develop the distribution maps of MC and a*value,which was also corresponding to the temperature distribution.(2)Microwave heating time had a significant influence on tenderness of beef,and the HSI was succussfully used to predict WBSF and protein of beef.The WBSF value increased obviously in two separate phased from 0-30 s,and again from 45-70 s,which was due to the different denaturation temperature of myofibrillar protein and collagen protein.For predicting WBSF value,the SNV-RC-LS-SVM model based on data fusion achieve best result with R_P~2of 0.911 and RMSEP of 4.491 N.For predicting myofibrillar protein,the SG-SPA-PLSR model based on data fusion showed best result with R_P~2 of 0.945 and RMSEP of 11.152 mg/g.(3)The doneness degree of beef during microwave heating was divided into 5 degrees,and the HSI was succussfully used to classify different doneness degree of beef.According to quality changes of beef during microwave heating,the doneness was divided into zero-class doneness(raw meat),the 1st-class doneness,the 2nd-class doneness,the 3rd-class doneness,and the 4th-class doneness.For doneness classifition based on spectral and texture information,two simplified models of Der2-RC-LS-SVM and Der2-SPA-LS-SVM based on data fusion have the best results with high correct classification rate of 93.0%and 95.0%,respectively.For doneness classifition based on doneness score equation,the classification model had the correct classification rate of 86.0%.All the above results demonstrated that the HSI technique was an effective method to distinguish between different beef doneness classes during microwave heating process.