Comparing performance of classification algorithms to use for grading coffee’s raw quality by using image processing techniques

Abstract

This study tries to apply digital image processing techniques towards sample Coffee raw quality value grading. More specifically, this study emphases on comparing performance of classification algorithms to use for grading coffee raw quality by using image processing methods. To ease experimentation image processing phases are followed, including image acquisition, image preprocessing (image filtering and attribute selection), image analysis (segmentation, feature extraction and classification), and image understanding for raw quality image grading. Artificial Neural Network, support vector machine and K-Nearest neighbor classifiers on each classification parameter of morphology, color and the mixture of the two has been made. Experimental outcomes confirm that Artificial Neural Network classifier generated the highest performance of 89.45% accuracy as compared to support vector machine (with 83.75%) and K-Nearest neighbor classifier (with 77.85%). Thus, suitable selection of image processing and classification techniques paves the way for higher accuracy in the higher-level process for decision making.