Archives of Current Research International

Traditional fruit grading relies heavily on manual inspection, which is labour-intensive, subjective, and prone to inconsistent decisions in post-harvest handling. This study designed and developed a low-cost fruit grading system using digital image processing in MATLAB integrated with an Arduino UNO-based sorting mechanism. The prototype consisted of a USB camera, controlled illumination, an IR sensor, a conveyor belt, an L293D motor driver, DC gear motors, and collection bins for graded fruits. Images captured in RGB format were processed in MATLAB through conversion to HSV colour space, Gaussian noise reduction, hue-based segmentation, morphological refinement, and feature extraction. The mean hue value, supported by saturation and area information where required, was used to classify fruits into ripe, unripe, and defective categories based on predefined thresholds. Classification results were transmitted to the Arduino through serial communication to actuate the sorting gate. The system was evaluated using 100 samples, including 40 ripe, 35 unripe, and 25 defective fruits from tomatoes, bananas, and mangoes. The prototype achieved an overall classification accuracy of 91.0%, with category-wise accuracies of 95.0% for ripe fruits, 91.4% for unripe fruits, and 84.0% for defective fruits. Processing time ranged from 0.8 to 1.2 s per fruit, supporting a throughput of approximately 60-70 fruits per minute under controlled conditions. The results indicate that MATLAB-based image processing integrated with Arduino control can support practical, repeatable, and cost-effective fruit grading for small- and medium-scale applications.