Face Recognition using Eigenfaces Method
| dc.contributor.advisor | Teferi, Dereje (PhD) | |
| dc.contributor.author | Arega, Daniel | |
| dc.date.accessioned | 2018-11-26T12:39:10Z | |
| dc.date.accessioned | 2023-11-29T04:56:47Z | |
| dc.date.available | 2018-11-26T12:39:10Z | |
| dc.date.available | 2023-11-29T04:56:47Z | |
| dc.date.issued | 2011-05 | |
| dc.description.abstract | This study attempted to develop a prototype face recognition system and the performance of the system is tested on a face database of Ethiopian faces. An Eigenfaces approach, which was one of first successful demonstrations of machine recognition of faces, is adopted. Towards this end, literature is reviewed on face recognition, Eigenface method and Principal Component Analysis (PCA). In the progression, a database of 76 face images, of 14 different individuals, was constructed. The database constitutes 14 normal frontal faces (Category_A), 14 face with a smiley expression (Category_B), 7 face images from Category_A and 7 from Category_B with head orientations rotated by 450, and 14 face images where the background in Category_A were removed by appropriate mask, 14 face images by decreasing the illumination level in Category_A by 10% and 6 more face images from Category_A after facial details are added manually. Appropriate tools, like Matlab development environment, were used to realize the system. The proposed system has four major components; the preprocessing module, feature extraction and dimension reduction via PCA module, database construction and updating module and face recognition module. The experimentation process involves determining critical threshold values’ the system uses in times of recognition. The system was tested to explore the impact of changes in head orientation, illumination level, and face background. Finally the performance of the system was tested and the preprocessing module was used to improve the accuracy. The result shows that the system performs very well for probes in the face library but the general performance is found to be 85.71%. Moreover when top five ranks are considered 92.86% accuracy was achieved. In conclusion, its observed that the eigenface algorithm performs well on a database of Ethiopian faces. The results are encouraging and with more optimization works, such as using face detection algorithms and construction of larger face databases, as per the recommendations made in the research work, better results can be achieved in the future. Keywords: Face Recognition, Eigenfaces Method, Principal Component Analysis (PCA) | en_US |
| dc.identifier.uri | http://etd.aau.edu.et/handle/123456789/14516 | |
| dc.language.iso | en | en_US |
| dc.publisher | Addis Ababa University | en_US |
| dc.subject | Face Recognition | en_US |
| dc.subject | Eigenfaces Method | en_US |
| dc.subject | Principal Component Analysis (PCA) | en_US |
| dc.title | Face Recognition using Eigenfaces Method | en_US |
| dc.type | Thesis | en_US |