COMPUTER APPLICATIONS ON PATTERN RECOGNITION FOR AGRICULTURAL PURPOSES

Abstract

In this project, several algorithms have been used and tested in order to obtain an accurate self-learning based pattern recognition applicationfor grading agricultural and food products.This application consisted of two parts. The first is the pattern recognition part. It dealt with processing the digital image. Each image was processed to obtain values for three main classification factors which are size, color, and texture. The digital image processing part processed, extracted, and saved the values. For each one of the three classification factors we implemented different algorithms and methods. The second part of this work is the classifier. Itused four different classifiers(Fuzzy logic, Artificial Neural Network, Support Vector Machine, and K-Nearest Neighbors) with accuracy 80%,90%,91.5%and 82.5%, respectively. This application used siwi date fruit and peanut as a variety to build and test the application. A total of 1033peanut samples, and 570 date fruit samples were used to build and train the system. The program developed has been able to distinguish the three different classes of palm date fruit and two different classes of peanut automatically. This study provides a very good technique to standardize the fruit grading system over a large area.