The objective of this project is to develop a robust automated algorithm for real time pupil and iris detection with high recognition rates in varying environment. First, Haar cascade based algorithm has been applied for fast and simple face detection from the input image. Then the eyes are extracted from the face and apply the algorithm for deciding left and right eye. After that, the pupil is extracted from both left and right image of eye by leveraging the property of black intensity distribution. Parallel, the iris candidates are extracted by using the new designed filter of size 5 X 5.
Costs of each iris candidates are calculated. Finally, the iris candidates are paired up and the cost of each possible pairing is computed by a combination of mathematical models. After that, these irises pair will be treat as information for system to continue the tracking it in the continuous frame. The novelty of this work is that the algorithm works on complex low resolution images without constraints on the background or surrounding lighting. This algorithm has achieved high success rate of automatically iris recognition when the system has more online based positive example that generated during the tracking process under user friendly real time environment.

Introduction

From a computer scientist’s aspect, human beings are machines which receive input from their sensors such as ears, eyes, skin and which interact with the world they live in through their actuators, which are their hands, feet, and so on. Therefore, as in the case of robots, one can understand the basis of their reasoning by inspecting the input they receive and also how they direct the attention of their sensors, for instance by looking at specific locations or inspecting unknown objects by touching or smelling.
Eye-tracking studies build upon this insight and analyse the relation between the movements of a person’s eyes and their attention, reasoning and feelings. These studies make use of special equipment called eye-trackers, which either calculate the direction of a person’s gaze or a point in the surrounding area where the gaze is fixed at. This point may be located on a 2D plane (i.e. the display of an electronic device) or a 3D volume and is also called the point of regard (PoR). There exist several commercial models of eye-trackers which come with their software platforms to carry out eye-tracking research, also expertise for calibration and actual usage is provided through documentation and support.
Although the commercial components provide satisfactory performance for marketing or scientific research, the scalability of this methodology is problematic because these products require trained operators and their price (starting from several thousands of Euros) makes them not suitable for this purpose. Therefore, the availability of a cheap, easy to setup alternative which does not require special hardware and which provides comparable performance is a necessity.

Building this alternative constitutes the problem we address in this work, and it consists of understanding where a subject is looking at using cheap components such as light sources, cameras and a computer to run the eye-tracker software. Actually, these requirements are already met in many consumer electronic devices such as laptops, smart-phones, tablet computers. We believe that the availability of a system running on these machines will provide basis for a variety of applications. For example, the eye-tracker can be used as a new input method for the electronic device just like a mouse, or it can be used as a tool to enable remote marketing and usability research. The system can replace commercial eye-tracking equipment to the extent allowed by its performance, and it can enable mobile applications of these techniques.