Facial recognition I
In this project, Principal Component Analysis (PCA), one of the methods in Facial Recognition will be discussed. The objectives of the project is to research on PCA, involving the understanding of the concept, implementing it with C++ language, conducting experiments and identifying its strengths an...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
2013
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/52624 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | In this project, Principal Component Analysis (PCA), one of the methods in Facial Recognition will be discussed. The objectives of the project is to research on PCA, involving the understanding of the concept, implementing it with C++ language, conducting experiments and identifying its strengths and weaknesses.
The project describes the mathematical theory of PCA. It also introduces an open-source library for the implementation of PCA using C++ language. Experiments were conducted to find the effects of number of eigenvectors; size of the training database; and threshold value has on facial recognition using PCA.
The results of the experiment showed that it is sufficient to use the first 10 eigenvectors for PCA to function efficiently. Furthermore, from a face database containing 40 subjects, using 6 out of 10 images per subject is enough to train the system with high recognition rate. The threshold value of 1.8 x 〖10〗^6 is the largest for the implemented program to achieve high percentage of matched image, low mismatched rate and yet capable to identify untrained subject effectively. |
|---|