Development and Performance Evaluation of Hausdorff Distance Algorithm Based Facial Recognition System
Securing access to information is of primary concern in many frame of reference including personal, commercial, governmental and military purpose. Computer verifiable biometric such as face provide an attractive means of securing access to information. Earlier algorithms for facial recognition system which includes Linear Discriminant Analysis (LDA), Principle Component Analysis (PCA) andIndependent Component Analysis (ICA) have yielded unsatisfactory result especially when confronted with unconstrained scenarios such as varying illumination, varying poses, expression and aging. This work presents a facial recognition authentication system using hausdorff distance algorithm in combating the highlighted problems. A system camera was employed for capturing images, information was stored using MYSQL database and biometric templates were stored as binary large object (BLOB). The developed system performance was evaluated using False Reject Rate (FRR), False Accept Rate (FAR), and Receiver Operating Characteristic Curve (ROC graph) as performance metrics. Tests were conducted at various threshold values. FRR errors obtained are 20%, 7%, and 2% at 500 threshold value for one-try, two-try and three-try configuration respectively. The system also presented FAR error of 0% at 500 threshold value for all configurations. As threshold value increases, FAR reduces while FRR increases.
- There are currently no refbacks.