Penentuan Jenis cacat Biji Kopi dengan Pengolahan Citra dan Artificial Neural Network

Abstract

Determination of the defect types of coffee bean is usually carried out visually by those persons having expertise due to experiences. This method is exhaustive and imprecise since it is influenced by human fatigue. The objective of this research is to identify the defect types of coffee bean by using a digital image processing technique and an artificial neural network (ANN). Coffee bean image was taken using a digital camera and then processed by an image processing program. Two ANN models were developed. The first model had 10 input parameters while the second model had five input parameters. Both models had altogether 26 output parameters of the defect types. The accuracy of the first model was 72.6% while the second one was 68.2%.