Color Face Recognition using Quaternionic Gabor Filters

While automated face recognition is a very important area of research and implementation, very few techniques make use of color information in recognition. However, both testing and analytic work indicate the potential advantages of using color. Existing work in Gabor filters for face recognition shows great promise, especially in robustness to illumination and pose variations. This proposed research will extend the well-known Gabor filters to color images using a specific form of hypercomplex numbers known as quaternions. More general methods of determining the locations on the face at which to apply the quaternionic Gabor filters, using statistical properties of a representative collection of face images, will be explored. The derived algorithms will be implemented and tested on a number of color face image databases.