To Instant of Coffee Beans using K-nearest Algorithm Over Clustering for Quality and Sorting Process

The proposed model articulated in this paper is a future mode of automated sorting of green coffee bean image with vision processing and machine learning. The presentations in the algorithm basically exploit visual characteristics of the bean, mainly characterized according to color, morphology, shape and size, in order to sort it using the KNN method. The entire method encapsulates capturing the images/source data-processing-feature extraction-classification. Essential image processing techniques include noise reduction, segmentation and others, to characterize and quantify necessary attributes of the bean. All characteristics will be feeding into the KNN classifier that determines granule quality and defect definition. MATLAB and Arduino-software in real-time decision making. Experimental mistakes of the whole system demonstrate that it has an improvement in the accuracy in identifying defects and quality assessment over conventional human inspection methods. The same model could increase efficiency, labor cost, and raise the entire quality control of procedure in the coffee field.