Application of Near Infrared Reflectance Spectroscopy to Cheese Analysis

The suitability of near infrared reflectance spectroscopy for determining cheese composition was investigated. The reflectance spectra of 30 hard and semihard cheese samples, including Cheddar, Colby, Parmesan, Romano, and Gruyere, were obtained in the region .9500 to 1.9500 μm. The same samples were freeze-dried, and their reflectance spectra determined in the region 1.050 to 2.3985 μm. Multiple linear regression analysis was used to select the best three wavelengths for predicting fat, protein, moisture, and moisture in nonfatty substance. The selected wavelengths for these constituents were in the 1.17 to 1.35 and the 1.64 to 1.71 μm regions of the spectrum. The correlation coefficients for reflectance with constituent concentration were .942 for fat, .938 for protein, .965 for moisture, and .975 for moisture in nonfatty substance. Coefficients of variation of the linear regression for the above correlations were 4.18, 5.23, 4.36, and 3.85%. The relationship of reflectance at selected wavelengths to constituent concentrations was linear over the entire concentration range. Concentration ranges were 34.6 to 56.7% for fat, 37.8 to 68.7% for protein, 15.8 to 41.7% for moisture, and 22.9 to 61.9% for moisture in nonfatty substance.