Study with the several resources on Docsity
Earn points by helping other students or get them with a premium plan
Prepare for your exams
Study with the several resources on Docsity
Earn points to download
Earn points by helping other students or get them with a premium plan
abstract of my master with research work thesis
Typology: Thesis
1 / 15
This page cannot be seen from the preview
Don't miss anything!
This Thesis with title PERFORMANCE EVALUATION OF SOME SELECTED FEATURE EXTRACTION ALGORITHMS IN EAR BIOMETRICS Submitted by ADEMILUYI, DESMOND TOYE Has satisfied the regulations governing the award of the degree of MASTER OF TECHNOLOGY OF LADOKE AKINTOLA UNIVERSITY OF TECHNOLOGY, OGBOMOSO, OYO STATE, NIGERIA.
…………………………………. Date:………………………… Dr. A. O Afolabi Supervisor
………………….……………… Date:……………………….. Dr. (Mrs) A. B Adetunji Head of Department
This work is dedicated to my Parents, respected Teachers and to all the people who still live in the Red brick house.
Ogunwale, Tobi and Ayo Maria for all their support in the collection of the biometric data needed for my research. Achievements in life do not come without the blessings of the Almighty, teachers and parents. I express my deep sense of respect and profound gratitude to my father and mother for all their efforts in raising me up to this stage and enabling me to devote myself to my studies without having any worries. My sincere gratitude to my brothers and sisters for their support, patience, understanding and encouragements. Finally, I will not forget to mention my little nephews: Sinaayo and Pelumi, my niece Sekemi and my little brother Debo whose loving presence, I always long for.
The features of the human ear are rich and stable but reported low accuracy of ear recognition systems as a result of feature extraction techniques employed has kept the system away from being widely used. Principal Component Analysis (PCA), Gabor Filter (GF), Geometric Method (GM) and Speeded Up Robust Features (SURF) are feature extraction algorithms that are commonly used in ear biometrics recognition. However, researchers have not determined the postures where each algorithm is most appropriate. This research work carried out performance evaluation of these four feature extraction algorithms to determine the appropriate algorithm for different postures. A database of 288 ear images of 20 x 20 pixels was created by acquiring three (right) ear images with normal pose from each of 72 individuals and a set of 72 ear images from another 72 subjects (imposters) using a 12- mega pixel digital camera. One hundred and forty four (144) images were used for training while the remaining 144 were used for testing. Image pre- processing was carried out by conversion of the ear images into grayscale, the grayscale images were then normalized to a constant mean and variance. PCA, GF, GM and SURF algorithms
earring while GM obtained best results for ear symmetry. This could serve as a guide for researchers to choose appropriate algorithm for ear features extraction.
Title Page i Certification ii Dedication iii Acknowledgement iv Abstract vi Table of Contents viii List of Tables xii List of Figures xiii CHAPTER ONE INTRODUCTION
4.7 Matching Speed and Memory usage 69 4.8 Overall Discussion 73
CHAPTER FIVE CONCLUSION AND RECOMMENDATION 5.1 Conclusion 76 5.2 Recommendations 77 5.3 Contributions to Knowledge 77 References 80 Appendix A: Recognition Time and Memory Usage across the Four Feature Extraction Algorithms, Part 1 82 Appendix B: Recognition Time and Memory Usage across the Four Feature Extraction Algorithms, Part 2 83 Appendix C: Recognition Time and Memory Usage across the Four Feature Extraction Algorithms, Part 3 84 Appendix D: Recognition Time and Memory Usage across the Four Feature Extraction Algorithms, Part 4 85 Appendix E: Recognition Time and Memory Usage across the Four Feature Extraction Algorithms, Part 5 86
4.1 Performance Analysis of Feature Extraction Algorithms for Normal Pose using 72 Genuine Subjects and 72 Imposters 54 4.2 Performance Analysis of Feature Extraction Algorithms for Slant Pose 55 4.3 Performance Analysis of Features Extraction Algorithms for Minor Occlusion with Earring 57 4.4 Performance Analysis of Feature Extraction Algorithms for Recognition Using the left Ear 58 4.5 Average Time and Memory Usage over the Four Feature Extraction Algorithms 59 4.6 Performance Comparison of Feature Extraction Algorithms 74
2.1 A Generic Biometric System 8 2.2 Anatomy of the Ear 10
