The Effect of Distance Measures on the Recognition Rates of PCA and LDA Based Facial Recognition

Many components affect the success of a facial recognition system. While some research attempts to improve on PCA or LDA algorithms, an often overlooked component is the distance measure. In this paper we show that the choice of distance measure greatly affects the recognition rate. Experiments are performed using the FRGC and FERET face databases. Recognition rates of ten distance measures are compared. There is an inconsistency of performance for each distance measure across each algorithm and face database. This shows that being able to determine the best distance measure before running the recognition algorithm will make the recognition system more successful.