Performance Evaluation ofFaceRecognition inthe

Inrecent years, manydifferent imagefeatures have beenusedforfacerecognition. TheGaborwavelet feature is themostwidely usedimagefeature infacerecognition systems because itsrecognition rateissuperior tothatofEigenface systems. However, itremains unclear astowhether Gaborwavelet features areindeed thebestwavelet features forfacerecognition. Inthis paper, weextract imagefeatures offacial images from various wavelet transforms (e.g., Haar, Frenchhat, Mexican hat, Daubechies, Coiflet, Symlet, and0-spline) andevaluate their facerecognition performance. We alsocomparetherecognition performance offixed- andadaptive-scale wavelet features. The results demonstrate thattheperformance ofthewavelets assessed hereissimilar tothatoftheGaborwavelet, andthatthe performance ofadaptive-scale wavelet features issuperior tothat offixed-scale features.