Effect of Wavelet Decomposition on Database Size Reduction for Face Recognition Rate

This paper presents the effect of the discrete wavelet decomposition on the database reduction, which is used in biometric face recognition. The Olivetti and Oracle Research Laboratory is own face databases for training and testing. The first level wavelet decomposition was applied to the training and testing faces made the size of database decreasing by 75% out of original size. Both databases are treated by principal component analysis algorithm to get the effect of that on face recognition rate. Principal component analysis algorithm is biometric face recognition. The experiment results appeared the reduction in face recognition rate between original and reduced database is 2%. This approach is more efficient for reduced the database, and computational load with a maintaining an acceptable proportion of recognition rate.