Modelling of apple slice moisture content by optical methods

Experiments were performed to follow the moisture loss of apple slice samples (discs of 3 cm diameter and 1 cm thickness) during the drying process. Different optical methods were tested in order to find a model for prediction of the moisture content based on non-contact measurements. Apple discs were dried in a hot air drying chamber for different periods (0–7 hours). The mass of every individual sample was measured before drying (initial mass), after drying (actual mass), and after the optical tests at the end of a 24 h drying process (final mass). Both the wet base and the dry base moisture content were calculated from the actual mass and the final mass of the samples. The optical properties of the samples of different moisture content were measured by two different optical methods. Laser induced backscattering method, applying 3 mW laser modules of different wavelengths, was used to generate the diffuse reflection pattern (halo) on the surface of samples and evaluate the halo properties with a machine vision system. Near Infrared Reflection (NIR) technique was also used to collect/measure the log(1/R) spectra of the samples in the 740–1700 nm range.PLS method with full cross-validation was used to predict the moisture content of the samples based on the backscattering data (quantitative parameters of the halo profile) and on the NIR spectra (raw and transformed log(1/R) data). Effective models (r>0.98, RPD>5) were found for prediction of the dry base moisture content of the samples based on both optical methods