CIEDE2000-based Wine Color Analysis Using Smartphones in Unconstrained Environments

Wine color offers key insights into composition, origin, and aging, serving as a crucial indicator for classification and quality assessment. Traditional methods, based on sensory evaluation and spectrophotometry, are often environmentally sensitive and expensive. This study explores digital image analysis and perceptual color similarity metrics to achieve coarse wine classification in unconstrained environments using smartphones. Through machine learning, we identify color centroids for red, white, and rosé wines, employing threshold-based, Support Vector Machine (SVM), and closest-centroid approaches. By leveraging the CIEDE2000 metric alongside other distance-based techniques, our SVM model achieves 98% classification accuracy (F1-score = 0.979). This demonstrates a cost-effective, accessible alternative to invasive laboratory methods for wine color analysis, even under diverse, unconstrained conditions.

This is the accepted version of the paper presented at EEITE 2025. The final published version is available on IEEE Xplore at 10.1109/EEITE65381.2025.11166185 .

ACKNOWLEDGEMENTS
This work was conducted as part of the Watson project, funded by the European Union's Horizon Europe research and innovation programme, under grant agreement No. 101084265.

DISCLAIMER
Funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Research Executive Agency (REA). Neither the European Union nor the granting authority can be held responsible for them.