Influencing wavelength comparison in pls models for the prediction of pork meat quality by vis-nirs

Antoine Vautier, 65th International Congress of Meat Science and Technology (ICOMST), 4-9 août 2019, Potsdam/Berlin, Allemagne, poster

For the needs of a running project dealing with hyperspectral imaging, we wanted to define the spectral range specifications of the device being built. The reference data we expect to predict on pork meat cover a wide diversity of parameters: pH, drip loss, cooking yield and intramuscular fat. In previous studies focusing on VIS-NIRS (Visible and Near Infra Red Spectroscopy) prediction of the meat quality, we mainly used the 350-1800 nm spectral range of a Labspec4 spectrometer (ASDI). Because this spectral range seems to be too wide for a single ready-to-use hyperspectral camera, we decided to take another look on our previous NIRS studies and investigate which wavelength are the most predictive. The objective was to make a comparison of the influencing wavelength that are included in the PLS prediction equations, for each of the meat quality parameters we had studied.