Firmness has been accepted as an important quality criterion in fruits and vegetables. However, firmness measurements applied on agricultural produce are destructive in nature and applied only on limited number of selected produce instead of applying them on all the products available. On the other hand, automated quality control systems have been aiming for non-destructive quality measurements on every single produce and for sending the product to the right market with a good competitive power after the evaluation. NIR spectroscopy has an important potential in this sense and recently, it has been studied intensely with the purpose of determining inner as well as outer quality criterions of agricultural produce non-destructively. With this study it was aimed to develop a firmness quality prediction model based on partial least squares (PLS) method to determine fruit flesh firmness of cherry tomatoes non-destructively using FT-NIR reflectance spectroscopy. For this purpose, first FT-NIR reflectance spectrums of cherry tomatoes, which were grown in a green house and harvested at 6 different maturity stages, were taken using a FT-NIR spectrometer (Bruker Optik, GmbH, Ettlingen, Germany) and then, firmness measurements were performed using TA-XTP Plustexture (Stable Micro Systems Ltd., UK) texture equipment and TextureExponent 32 software. After these measurements, chemometrics relations were sought between FT-NIR spectrometer readings and firmness measurements (Force-1, Area, Force-2) based on partial least squares (PLS) method using The Unsrambler X (Version 10.5.1, Camo A/s, Oslo, Norway) software for the purpose of developing firmness prediction models. In the results, fairly good prediction models were obtained between FT-NIR reflectance spectrums and firmness measurements. According the prediction results, Force-1 firmness value was predicted with a coefficient of determination of 0.77 (RMSEP= 251.01) (R2=0.75, RMSEC= 239.56 in calibration) while Force-2 and Area firmness criterions were predicted with coefficients of determination of 0.78 (with RMSEPs of 276.98 and 427.39, respectively) (with R2 values of 0.76 and 0.75 and, RMSECs of 266.25 and 412.37, respectively). These results can be assumed as good prediction results considering that the investigated quality criterion is firmness, which is hard to predict non-destructively.
Anahtar Kelimeler: FT-NIR Reflectance Spectroscopy, PLS Analysis, Cherry Tomato, Non-destructive firmness Prediction