Automated guava classification technique using artificial neural network and artificial ant colony raed qaqish ORCID: 0000-0003-2317-6183

Abstract

Guava is one of the main agricultural products in the northern part of Jordan. The classification of fruit products is an essential part in the packaging, shipment, and marketing processes. Normally, the classification and grading of fruits are performed manually, which incur additional overheads in the harvesting and marketing procedures. This adds to the final pricing of fruits and vegetables. The automatic classification and grading of fruits and vegetables can save substantial costs and efforts in addition to avoiding long delays. In this work, we propose an automatic classification system for grading and classifying guava fruits using image-processing techniques. The proposed research will provide a rich analysis and investigation of several features of the problem at hand. Our study will convey and integrate shape, color, and texture descriptors. During the preprocessing step, many morphological operations will be applied accompanied with filtering using various filters like the Wiener Filter. In the classification phase of the proposed system, at least two classification approaches will be considered including the artificial ant colony algorithm and the minimum distance classifier. Moreover, in the experimental part we will collect a large number of input samples (images) with different projections.