Raw Quality Value Classification Of Ethiopian Coffee Using Image Processing Techniques In The Case Of Wollega Region

Development of an automated computer vision system aiming in the establishment ofrntechnological and innovative approaches towards sample coffee bean raw quality valuernclassification by extracting the relevant coffee bean features is the focal issue of thisrnexploratory research. Of paramount significance in this regard is addressing the identifiedrnproblems of the tedious and inefficient manual grading and sorting mechanisms of one ofrnthe most important agricultural products in Ethiopia, coffee. Prevalent sorting andrnclassification approaches are characterized by subjective assessments of the features andrnnature of this huge economy representing crop, thereby influencing quality control andrnproductivity aspects of the product. The major objective of the research spans extractionrnand selection of the important coffee bean morphological and color features that arernuseful for the purpose of classification of the raw quality grade level of sample coffeernbeans by designing, analyzing and testing a digital image processing model.rnThe automated raw quality value classification experimentation comprised the analysis ofrnimages of washed coffee beans of varying grades from Wollega region, using majorrnattributes of morphological structures (shape and size), and color features. Grades 2 – 9rnof the coffee beans were available, providing a total of 27 samples, which yielded 324rnsample images after a series of re-sampling measures of same into 12 sub-samples. Thernoverall image processing work to develop models and depict trends for an efficient rawrnquality value classification involved sequential phases of image acquisition, imagernenhancement and segmentation, feature extraction, attribute selection, classification andrnperformance evaluation.rnThe Naïve Bayes, C4.5 and Artificial neural networks (ANN) were implemented for suchrnclassification purposes. A combined morphological and color features aggregate functionrndataset was used to develop the base model, though model attempts with separate featuresrnwere conducted. Feed-forward multilayer perceptrons with two hidden layer and backpropagationrnalgorithms are used in the ANN classifiers.rn rn rnDiscretization of the raw quality value in to three interval classes was done to improvernthe performance of the model. 75% split evaluation technique was implemented for thernNaïve Bayes and ANN classifiers as 10-foldcross validation evaluation techniquesrnimplemented in C4.5. Naïve Bayes classifier yielded higher model performance (82.72%rncorrectly classified), followed by C4.5 (82.09%) and the ANN classifier (80.25%). Modelrnrobustness and sensitivity was analyzed by using perturbation analysis involvingrnmanipulations of model evaluation techniques and dataset characters. Alteration ofrnnumber of beans in discretization and the use of different number of hidden layersrnconstitute the trial modeling in this regard. Classification model was also run with variousrncombinations of features of the coffee beans as listed with the attribute selection featurernof Weka tool, where the final selection of the 21 features was done at a maximal modelrnperformance level for the Naïve Bayes and ANN classification approachs. C4.5 selectedrn10 features as it has its own attribute selection characteristics.rnAn additional simulation was done with regression analysis for the sake of evaluation andrntrends analysis of the model outputs. A higher relation was resulted from this statisticalrnapproach between the raw quality values and the mentioned coffee bean features,rnsupporting suitability and accuracy of dataset for classification in this research.

Get Full Work

Report copyright infringement or plagiarism