M.Sc. Tezi Görüntüleme

Student:	Tuğrul KARAKAYA
Supervisor:	Assoc. Prof. Dr. Vasif Vagifoğlu NABİYEV
Department:	Computer Engineering
Institution:	Graduate School of Natural and Applied Sciences
University:	Karadeniz Technical University, Turkey

Title of the Thesis:	FACE DETECTION WITH NEURAL NETWORKS
Level:	M.Sc.
Acceptance Date:	4/7/2005
Number of Pages:	122
Registration Number:	i1598

Summary:

The thesis deals primarily with the face recognition and endeavors to identify the required invariants to detect the human face. The study explored the details related to the face detection needed to recognize the pattern of face. The study generated various methods for enabling much faster process of face detection.

The study mainly focused on the use of neural network for pattern recognition. Face detection was performed by multi layer neural networks that use back propagation algorithm as training method. These efforts revealed that neural networks developed for face detection can easily be adapted to other similar applications. 15 different posturing images taken for 50 people were used to test the systems. Database with 750 images included the different and challenging postures such as wearing glasses, shawl, left-sided and right sided. Finally, an improved application that can recognize faces from the front for different images with varying background and illumination was created.

Keywords: Face recognition, Face detection, Neural networks, Pattern recognition, Image processing, Biometrics.