<![CDATA[The citrus farming industry faces major challenges in maintaining product quality consistency due to subjective manual sorting processes that are prone to fatigue and have varying standards. This problem results in economic losses due to errors in detecting ripeness levels and physical damage that hinders market competitiveness. This study aims to design and implement an automated citrus fruit quality evaluation system using a real-time X-Means algorithm. The research method begins with visual data acquisition through a camera sensor using the automatic snapshot feature to convert physical objects into digital data. The data then undergoes preprocessing, which includes filtering to remove noise, color (RGB) and texture feature extraction, and normalization using Min-Max Scaling to balance parameter weights. The X-Means algorithm is used because of its ability to independently determine the optimal number of clusters through the evaluation of the Bayesian Information Criterion (BIC) score. The processing results show that the system is able to accurately group oranges into three categories: ripe, which are dominated by bright orange colors; unripe, which are dominated by green colors; and rotten, which are identified through rough textures and dull colors. The integration of this technology ensures that all decision-making occurs quickly and objectively, providing a practical solution for the industry to consistently improve product quality control efficiency in the field.]]>
Copyrights © 2026
