Study On The Breeddiscrimination And Quality Determination Of Lamb By Near Infrared Spectroscopy

Although the yield of lamb in China was the first in the world, the best lamb didn't sell the highestprice in the meat market. There were different variety of breeds and grade of lamb mixed together. Evensometimes butcher sold shoddy lamb to customers, so how to determinate the quality of lamb was aburning question. Near infrared spectroscopy was one of the rapid development of high technology inrecent years. It was widely used in the food composition detection because near infrared spectroscopywas rapid, accurate, undamaged etc. In the research physicochemical quality of lamb was determinatedaccording to the Chinese standard, and then the lamb was scanned near infrared spectroscopy in order toestablish near infrared spectroscopy prediction model. At the same time establish the near infraredspectroscopy discrimination model of different breeds lamb. Specific results were as follows,It was analyzed about the lambphysicochemical qualities in the major part of China.Tibetanlambwas rich in protein (24.18%) and intramuscular fat(2.05%).The forelegs and hind legs hadsimilarphysicochemical composition, which were different in moisture content, protein content and fatcontent compared with neck and longissimus dorsum. The sexesinfluenced the lamb quality of Dorpersheep most among the test breeds, and it influencedthe physicochemical quality of Tibetan sheepslightest. Tan sheepwas influenced seriously by age, but Sunit sheep,Ujmuqin×Little Han Tailedsheepand Jinzhong sheep werethe least affected by age.Gathered the spectrum of NIRS and detected the physicochemical properties, the near infraredspectroscopy models were founded to predict the lamb chemical composition. The squares of r (R2)were0.87(water),0.72(protein),0.61(fat),0.65(water-holding capacity),0.44(ash),0.58(pH),0.59(shearvalue) and0.69(color), respectively.The root square error of cross validation was0.77(water),0.85(protein),0.47(fat),1.86(water-hold capacity),0.12(ash),0.26(pH),10.75(shear force) and2.05(color), respectively.The result showed that near infrared spectrum can be used to predict the watercontent, protein content, fat content, water-hold capacity and color oflamb.Establish the near infrared spectroscopy discriminant model by Linear Discriminant Analysis andPartial Least Square.The correct rate of the model established by LDA to training set samples was78.4%, and the cross validation accuracy was75.4%which means that NIR didn't recognize the9breeds lamb at the same time. Use the PLS to establish nine breeds lamb synchronous recognitionmodel and the highest correct recognition rate was48%. Set up single breed lamb discrimination modelby PLS and get higher correct rate. When the thresholding was set for±0.5, the correct rate was100%;when the thresholding was set for±0.3, the correct rate was higher than80%. So the NIR can be used forbreeds lamb discrimination.