Iris Recognition from 1-2 Meters
The goal of this project is to have a computer recognize a person from an image of his or her iris. The main application for iris recognition is in security, for example, access control. Iris recognition has been shown in independent studies to be the most accurate biometric. The drawback is that the iris is difficult to image. It is small and sometimes partly occluded by the eyelid and eyelashes. Current iris recognition systems require careful cooperation by the user and require the user's eye to be within about 10 inches of the camera. One goal of this project is to design hardware that can capture a good image of a person's iris from a distance of 1 to 2 meters with minimal user cooperation. This would overcome the inconvenience problem which is the main drawback of iris recognition technology. The major developments over last year are the completion of a novel segmentation algorithm and a novel comparison algorithm based on a new set of simple features.
Background & Objective: The problem of iris recognition has been studied by a number of researchers. Our objective is to develop a state of the art system. We are focusing on enabling iris recognition from further away than current products which will make it much easier to use. To do this we are investigating the hardware needed (cameras, lenses, pan-tilt units) as well as novel algorithms for comparing two iris images.
Technical Discussion: We intend to use MERL's state-of-the-art face detection and eye detection technology to enable a camera to be aimed at the iris from up to 2 meters away. With the right camera and lens, a high resolution image of the eye region can be captured at that distance. Once the eye region is captured, the iris must be segmented out. Given a segmented iris image, novel image features are computed to create a feature vector. Feature vectors can then be compared to yield a similarity value which is used to determine a match. We are investigating novel segmentation algorithms as well as novel features to use for comparison.
| Technical Reports: | |
| A System for Automatic Iris Capturing | |
Technology Area: Computer Vision
Modification Date: September 12, 2007