Next: BackgroundFacial Recognition Preface
Although humans have an amazing ability to distinguish and recognize faces, facial recognition on computers is an advanced field of study yet to be perfected. Only within the past decade have people seen the rise of face tracking in digital cameras and public security systems. Unlike humans, computers are easily confused by changes in illumination, variation in face angles, and accessories such as hair, glasses, and hats. However, computer face recognition is worth pursuing, because of the wealth of applications in security, digital photography, social networking, and other fields. This project uses the eigenface method to identify faces from still images.
 |
Facial Recognition Approach
When considering the proper approach for identifying faces, it was decided that the approach to be able to deal with imperfect settings such as different relative positions, sizes, and shapes of facial features such as the eyes, nose, cheekbones, and jaw. While facial features are a major component of some facial recognition algorithms, the extraction of landmarks were determined to be insufficiently robust for this project without a powerful normalization algorithm capable of features such as keeping a constant light intensity, correction of angles and pose variations, and accounting for accessories like glasses.
