Strategy for improving the reliability in the facial identification
Carlos M. Travieso, Jesús B. Alonso, Miguel A. Ferrer · 2005
This paper presents a simple, robust and novel for errors detection in biometric system which is applied to the Olivetti Research Laboratory (ORL) face database (400 images). We have used as parameterisation different transformed dominions (Travieso et al., 2004; Faundez, 2003), and a support vector machine (SVM) (Burges, 1998; Cristianini and Shawe-Taylor, 2000) as classifier. This system has been adjusted with our experiments for obtaining a false identification rate (FIR) of 0%, with a success rate of 90.8% a rejected samples rate of 9.2%.