Saturday 23rd of September 2017
 

A Hybrid Memetic Algorithm (Genetic Algorithm and Great Deluge Local Search) With Back-Propagation Classifier for Fish Recognition


Usama A. Badawi and Mutasem Khalil Sari Alsmadi

The aim of this study is to establish a hybrid method to optimize the performance of back-propagation classifier for fish classification by using Memetic Algorithm (MA) (genetic algorithm and great deluge local search). This is to be performed by utilizing the ability of memetic algorithm to optimize the parameters (weight and bias) of the back-propagation classifier (PBC). Recognizing an isolated pattern of interest (fish) in the image is based on robust features extraction. These features are extracted based on color signature measurements that are extracted by Gray histogram technique and Level Co-Occurrence Matrix (GLCM) method. The typical Back Propagation Classifier (BPC) has the slow practice speed and easy for running into local minimum disadvantages. The new system prototype will help in resolving such disadvantages. Results: The new system starts by acquiring an image containing pattern of fish, then the image features extraction is performed relying on color signature measurements. The system has been applied on 20 different fish families, each family has a different number of fish types and the used sample consists of 610 distinct fish images. These images are divided into two datasets: 410 training images and 200 testing images. The hybrid memetic algorithm(Genetic algorithm and Great Deluge Local Search) with back-propagation classifier (HGAGD-BPC) has outperformed the BPC method and previous methodologies by obtaining better quality results but with a high cost of computational time compared to the BPC method. The overall accuracy obtained using the traditional BPC was 84%, while the overall accuracy obtained by the HGAGD-BPC was 93.5% on the test dataset used. Conclusion: A powerful classifier for fish images classification has been developed. The new hybrid classifier is successfully designed and implemented and has performed efficiently to make a decision without any problems. Eventually, the classifier is able to categorize the given fish into its cluster (poison or non-poison fish) and categorizes it into its family.

Keywords: Back-propagation classifier (BPC), a hybrid memetic algorithm with back-propagation classifier (HGAGD-BPC), Color Histogram Technique, Gray Level Co-Occurrence Matrix (GLCM), Color signature measurements, digital fish images, poison and non-poison fish.

Download Full-Text


ABOUT THE AUTHORS

Usama A. Badawi
Dr. Usama A. Badawi has been graduated from the Faculty of Science Cairo university - Egypt in 1991, and has worked as a demonstrator in the same faculty. In 1995, he has finished his master degree in the field of object oriented databases. In the same year, he has worked as a lecturer assistant in his department. In the period from 1998 to 2001, he has got a scholarship from the DAAD to complete his Ph.D. research in the technical university of Darmstadt Germany as a visitor researcher at the computer science institute In the year 2001, he has finished his Ph.D. in the field of distributed systems and worked as a lecturer in the faculty of science Cairo University. Since 2004, he is working as an assistant professor at Dammam University Kingdom of Saudi Arabia

Mutasem Khalil Sari Alsmadi
MIS Department, Collage of Applied Studies and Community Service, Dammam University, KSA,


IJCSI Published Papers Indexed By:

 

 

 

 
About IJCSI

IJCSI is a refereed open access international journal for scientific papers dealing in all areas of computer science research...

Learn more »
Join Us
FAQs

Read the most frequently asked questions about IJCSI.

Frequently Asked Questions (FAQs) »
Get in touch

Phone: +230 911 5482
Email: info@ijcsi.org

More contact details »