Translation based Face Recognition using Fusion of LL and SV Coefficients

Abstract

The face is a physiological trait used to identify a person effectively for various biometric applications. In this paper we propose Translation based Face Recognition using Fusion of LL and SV coefficients. The novel concept of translating many sample images of a single person into one sample per person is introduced. The face database images are preprocessed using Gaussian filter and DWT to generate LL coefficients. The support vectors (SV) are obtained from support vector machine (SVM) for LL coefficients. The LL and SVs are fused using arithmetic addition to generate final features. The face database and test face image features are compared using Euclidean Distance (ED) to compute the performance parameters.