Klasifikasi Tingkat Kematangan Buah Salak Sidempuan Menggunakan K-Nearest Neighbor pada Ruang Warna HSV

Abstract

Determining the ripeness level of Salak Sidempuan fruit manually is often inaccurate, as it relies on human visual perception, which can be affected by lighting conditions, eye fatigue, and subjective judgment. Such limitations may reduce sorting quality and the market value of the fruit. This study proposes an automated ripeness classification system for Salak Sidempuan based on digital image processing using the K-Nearest Neighbor (K-NN) algorithm combined with the Hue Saturation Value (HSV) color model. The dataset consists of images of Salak Sidempuan at three ripeness stages (unripe, half-ripe, and ripe) captured using a smartphone camera. The process involves preprocessing, converting RGB images to HSV, extracting the average H, S, and V values, and classifying them using K-NN with Euclidean distance and K=5. The aim of this research is to design and implement a MATLAB-based application capable of classifying the ripeness level of Salak Sidempuan fruit quickly, accurately, and consistently. The evaluation was conducted on 15 test images to measure system performance using accuracy, precision, recall, and F1-score metrics. The experimental results show that the system achieved an accuracy of 80%, with high precision and recall values for the unripe and ripe classes. The Hue component proved to be the most stable parameter for distinguishing ripeness levels compared to Saturation and Value. This system is considered effective in supporting the objective sorting process of Salak Sidempuan and has the potential for further development into field-deployable solutions such as Android or IoT-based applications.