Assessment of Tomato Quality through Near-Infrared Spectroscopy—Advantages, Limitations, and Integration with Multivariate Analysis Techniques

This study conducted a systematic analysis and investigation of near-infrared (NIR) spectroscopy for assessing tomato fruits' quality. This was combined with the integration of multivariate techniques for analysis to improve existing methods for nondestructive quality assessment of tomato fruits. Variations in tomato characteristics were revealed by selecting spectral indices, leading to changes in reflection spectra at different wavelengths. To achieve more accurate predictions, the use of vectors containing various spectral indices became necessary. This study found that predicting key tomato characteristics relies more on the informativeness and predictive ability of vectors derived from spectral indices than on the method used to reduce data volume. Regression models were developed to automate the prediction of six tomato characteristics based on the spectral data. The developed models for vitamin C content, titratable organic acids, and lycopene showed the highest predictive efficiency (over 90%) among the investigated features. These results highlight the potential of regression models to be integrated into automated systems for assessing the quality of tomato fruits, providing sufficiently accurate information to enhance quality control processes.