RED KIDNEY BEAN (RKB) CLASSIFICATION AND GRADING USING IMAGE PROCESSING TECHNIQUES

Abstract

The red kidney bean (RKB) is a vital crop whose distribution in the market is subject to stringent quality control. RKB samples are now manually evaluated using ocular inspection, with the contents classified as foreign matter, defect, healthy, contrast, and insect board kernels. Visual examination, on the other hand, necessitates a significant amount of time as well as the presence of qualified and experienced professionals. Furthermore, it is influenced by human nature's biases and inconsistencies. Such a procedure cannot be adequate for large-scale examination and grading unless it is fully automated. The goal of this study is to create a system that can evaluate the quality of RKB sample elements utilizing digital image processing techniques, RKB image data is collected from ECX warehouse, the sample of RKB providing a total of 62 samples, which yielded 582 sample images. Image preprocessing are the steps taken for the improvement of the image data that suppresses undesired distortions or enhances some image features relevant for further processing and analysis task then a novel segmentation technique is proposed to segment the foreground from the background, partitioning both RKB and foreign particles and lay the foundation for feature extraction. To model RKB sample ingredients, a total of 24 features (14 colors, 8 shapes, and 2 sizes) have been extracted. The data set is randomly apportioned into training and test set with 70% and 30% proportion, respectively. Classification algorithms, such as artificial neural networks and naïve bayes classifiers are applied based on the Ethiopian Commodity Exchange (ECX) RKB standard. Using a feed-forward artificial neural network classifier with a back propagation learning algorithm, 24 input nodes, and 5 output nodes, matching the number of features and classes, has been constructed for the classification of RKB samples. Accordingly, the classifier achieved an overall classification accuracy of 93.8%. The success rates for detecting foreign object, defect, healthy, insect board, contrast, kernels are 100%, 92%, 95.2%, 84.4% and100% respectively. This research work does not include moisture content analysis of RKB. It is therefore recommended as a future research direction to enhance the performance of the proposed model in this study.