Automatic classification of physical defects in green coffee beans using CGLCM and SVM

Author

Montes Condori, Rayner H. ; Chuctaya Humari, Juan H. ; Portugal-Zambrano, Christian E. ; Gutierrez-Caceres, Juan C. ; Beltran-Castanon, Cesar A.

Author_Institution

Catedra Concytec en Tecnol. de la Informacion, Univ. Nac. de San Agustin, Areauira, Peru

fYear

2014

fDate

15-19 Sept. 2014

Firstpage

1

Lastpage

9

Abstract

This work is focused on the evaluation of physical coffee beans through a model of automatic classification of defects. The model uses a segmentation step that discriminates the background from the coffee bean image with a follow contours algorithm, then a CGLCM is introduced as features extractor and a Support Vector Machine for the classification task, a database of images has been collected with a total of 3367 images, the classification process used twelve categories of defects, the results of classification showed a accuracy of 86%. Finally a set of conclusions and future works are presented.

Keywords

agricultural products; image classification; inspection; matrix algebra; support vector machines; CGLCM; SVM; classification task; coffee bean image; gray level co-occurrence matrix; green coffee beans; image database; physical coffee bean evaluation; physical defect classification; support vector machines; Color; Electronic mail; Feature extraction; Image segmentation; Irrigation; Laboratories; Support vector machines; coffee bean; computer vision; feature extraction; segmentation;

fLanguage

English

Publisher

ieee

Conference_Titel

Computing Conference (CLEI), 2014 XL Latin American

Conference_Location

Montevideo

Type

conf

DOI

10.1109/CLEI.2014.6965169

Filename

6965169