<![CDATA[Potatoes are an important agricultural commodity whose quality directly affects market value and industrial processing efficiency. However, potato quality assessment is still dominated by manual methods, which lead to subjective assessments, inconsistent results, and limited processing capacity on a large production scale. This study aims to develop an automatic potato quality classification system based on digital image processing and Artificial Neural Networks (ANN) to provide an objective and efficient assessment method. The proposed system consists of six stages: image acquisition, pre-processing, segmentation using Otsu's threshold, morphological operations, feature extraction, and classification. A total of 272 potato images were used, consisting of 136 good-quality samples and 136 poor-quality samples, which were taken under controlled lighting conditions. Color features in the RGB color space and texture features extracted using the Gray-Level Co-occurrence Matrix (GLCM) were used as input features. The classification process used a Feedforward Backpropagation Neural Network with two hidden layers. The experimental results showed that the combination of RGB color features and GLCM texture features achieved the best performance, with an accuracy of 98.33% on the test dataset and an average testing time of 0.299 seconds per image. These results indicate that the proposed system can significantly reduce subjectivity and processing time in potato quality assessment, thereby supporting the implementation of a reliable and scalable automatic assessment system for agricultural and industrial applications.]]>
