Improved Pdf Based Face Recognition Using Data Fusion

In this paper a high performance face recognition system, based on different data fusion techniques combining the decisions obtained from the probability distribution functions (PDF) based face recognition system in different colour channels, is introduced. The PDFs of the equalized and segmented face images are used as statistical feature vectors for the recognition of faces by minimizing the KullbackLeibler Distance (KLD) between the PDF of a given face and the PDFs of faces in the database. Well known data fusion techniques such as Median Rule, Sum Rule, Max Rule, Product Rule, Majority Voting (MV) and feature vector fusion (FVF) have been employed to increase the recognition performance. The proposed system has been tested on the FERET, Head Pose, Essex University, and Georgia Tech University face databases. The overall results indicate that the median rule over-performs the other fusion techniques.