Hardware approach of ANN based iris recognition for real-time biometric identification

Abstract

Artificial Neural Networks (ANN) are increasingly applied to biometric identification because neural nets have been shown to be technologically powerful and flexible, ideally suited to perform identification analysis. Therefore, it demands the development of a new processing structure that allows efficient hardware implementation of the neural networks mechanism. This research presents the ANN based iris recognition for biometric identification modeled by the very high speed integrated circuit Hardware Description Language (VHDL) to ease the description, verification, simulation and hardware realization of this kind of systems. The project is divided into two processes which are image processing and recognition. Image processing was performed by using Matlab where back propagation was used for recognition. The iris recognition architecture comprises of three layers: Input layer with three neurons, hidden layer with two neurons and output layer with one neuron. Sigmoid transfer function is used for both hidden layer and output layer neurons. Neuron of each layer is modeled individually using VHDL. Functional simulations were commenced to verify the functionality and performance of the individual modules and the system. Iris vector from captured human iris has been used to validate the effectiveness of the model. An accuracy of 88.6% is achieved in recognizing the sample of 100 data of irises.