Eye-Identification System Based on Back-Propagation NN Recognizer

Abstract

This research paper deals with the implementation of an eye recognition system using neural network (recognition classifier). The proposed system contains two phases, preprocessing and recognition. The resized image is providing faster processing for training and testing, the size of image is 120×90 pixels. The preprocessed phase extract features from eye image and use it as input to neural network which uses back-propagation algorithm to recognize the eyes. The proposed algorithm use single neural network as classifier, which consists of three layers with tangent sigmoid, and linear transfer function respectively. The system trained 60 eye samples. After testing the system with (100) eye samples a recognition rate of 100% is obtained. This recognition rate value is perfectly suitable for eye recognition systems.

Keywords

• eye recognition
• neural network
• back-propagation
• pre-processing