The present study is concerning qualitative and quantitative detection of minced pork quality based on FT-near infrared (FT-NIR) spectroscopy and achieving the rapid approach to detecting the minced pork quality. Firstly, FT-NIR spectroscopy combined with partial least squares (PLS) and least squares-support vector machine (LS-SVM) was used for minced pork quality prediction including discrimination of the different muscle type of pig and quantitative detection of the fat, protein and moisture content of pork. The result indicated that 100% recognition ratio for calibration and 96% recognition ratio for validation were achieved by PLSDA for 4 different muscle types of pig. These two methods for chemical composition detection both have good performances in predicting fat and moisture content, the correlation coefficient for calibration and validation was all more than 0.9, but the models for protein content prediction were of less well performances, the correlation coefficients for calibration and validation, RMSEC, RMSEP and RMSECV respectively were 0.722, 0.593, 1.595, 1.550 and 1.888, respectively. The LS-SVM method is more accurate in predicting each quality index than the PLSR method. The result shows that the prediction models for fat and moisture content based on LS-SVM have a better performance with high precision, good stability and adaptability and can be used to predict the fat and moisture content of minced pork rapidly, and provide a fast approach to discrimination of the different muscle type of pig.