A Biometric Authentication Approach using Face Recognition System

Today, face recognition is extensively used in Computer Vision Applications such as for recognition as well as authentication purpose. The main backbone of the algorithm is to detect the face portion from the entire image, and for detecting the face, we have to detect the skin first. Conducting our research in various color transformation algorithm, it is found that the best method is to transform the RGB image into the YCbCr image to detect the skin color of humans by setting appropriate thresholds. The reason for using this color model is to remove the illumination component, Y. We have proposed and implemented a method to locate the face portion of human by determining the number of holes in the skin region using Euler's method and also by taking eye distance. Once the face is detected, these are stored as database. We use principal component analysis for the recognition purpose. The result of this entire system is shown in the figures and also their accuracies.