Prediction of Meat Spectral Signatures in the Near Infrared Region Using Optical Properties of Main Chromophores

Abstract. An approximation model for predicting spectral signatures of minced meat samples was developed depending on the concentrations and optical properties of the major chemical constituents. Minced beef samples of different compositions scanned on a near infrared (NIR) spectroscopy and on a hyperspectral imaging system were examined. Chemical composition determined heuristically and optical properties collected from authenticated references were modeled to approximate meat' spectral signatures. By assuming homogeneous distributions of the main chromophores in the mince samples, the obtained absorption spectra are found to be a linear combination of the absorption spectra of the major chromophores present in the sample. Results revealed that the developed models were good enough to derive spectral signatures of minced meat samples with a reasonable level of robustness and agreement index value more than 0.90 and ratio of performance to deviation (RPD) more than 1.4.