An Effective Multi-domain Facial Feature Based Improved FRS
Pushpa Sherlin, Suganthi Bai · 2015
Biometrics is a readily available metric that is playing a vital role in face recognition. There are many domain areas for the features extractionsuch as frequency measures, geometrical measures and statistical measures. In the proposed Face Recognition System (FRS)the extraction of features are done using frequency and statistics measures. The FFT,DCT, variation, standard deviation, mean, median and entropy are the five features extracted from the ORL dataset for recognition process. A local normalization technique is applied to the extracted features. The training and testing are performed with normalized features using Back PropagationNeural Network (BPNN).