3 / 2018-12-27 00:56:00

Features Extraction of Hybrid Algorithms (KL)and (ENN) As Biometric Identifier for Face Recognition

Facial recognition identity, principle component analysis, FAR, GAR, Biometric recognition

This paper offers a suggested method of facial recognition to identify any person. Initially, captured images in the database to apply matching processing, which ease the recognition operation. The most difficult problems in the recognition algorithms are, the difference in face appearance, the lighting effect, and the composite background of the image. One of the most necessary and effective applications suitable for biometric systems and image processing. This paper presents the fundamental step in the extraction of the properties that depend on the Principal Component Analysis (PCA) also known as Karhunen–Loe`ve(KL) to convert Eigen values to reduce the number of inputs on the network. To identify a person's image apply Artificial Neural Networks (ANN) using Elman neural work.
Practically taking photos (10) each person has eight poses (4 to 4 training to test). The experimental results of the ENN classification are calculated as a true acceptable rate (GAR) of 97% while the false acceptance rate FAR is equal to 3%.
In addition, the artificial neural network gave a good performance to anyone.