An embedded iris recognizer for portable and mobile devices

Software-intensive systems play an increasingly dominant role in our lives and daily activities. Several applications, in which a timely response to user and environment stimulus is essential, require real-time software intensive systems. Computation-intensive applications, such as video compression, control systems, security systems, result in significant growth for processor workload. To address the above issues, one possible solution is to design embedded specialized components. At the same time, the integration of new features in portable and mobile devices is rapidly increasing. Several services and applications require robust user authentication for access to services, data, and resources. In this work, an embedded iris-based recognizer is proposed to overcome some of the most common limits of real-time software-intensive systems such as response time, used resources and power consumption. Starting from the ophthalmologic study on iris micro-features, user authentication is based on nucleus and collarette minutiae. The whole recognizer has been prototyped through the Celoxica RC203E board, equipped with a Xilinx 3M VirtexII FPGA, using the Handel-C algorithmic-like hardware programming language. Experimental tests, performed using the widely used CASIA database, have shown FAR (False Acceptance Rate) = 0% and FRR (False Rejection Rate) = 8%, while the whole estimated iris processing and recognition process requires 139.05 mW (logic cell power consumption).