Iris recognition is likely to meet many challenges in biometrics. One of the problems related to the iris recognition involves recording images with high resolution. We admit that it takes between 100 and 200 points in the diameter of the iris, that means a resolution of about 50 to 100 microns. Depending on the application, it is planned to acquire iris images of people walking (for a portico at an airport) or just standing in front of the device - typically at one meter - (case of bank ATM).
The use of an adaptive optics system, ensuring focus and image stability, greatly improves the probability of success of this recognition. In the case of a walkng person, a tracking system with an adapted dynamic for the movement and a sufficient precision to follow the head must be coupled to the high resolution images camera.
Iris recognition systems on stopped people are already commercially available, such as that proposed by the company AOptix but they are expensive and bulky. In open literature, laboratory devices equipped with a tracking system and capable of making iris recognition on a moving person are presented. However, when performance of these systems is published, they are systematically estimated on good health and fully cooperative people. Public use requires a running on a representative sample of all population groups. However, movements of the iris (amplitude, frequency) seem influenced by the status of the person, much more critical especially in older people or those under stress for example. No communication reports the behavior of iris recognition systems based on the audience.
The objective of the proposal is to raise sticking points for the development of a low cost, compact and robust adaptive optics system, coupled to the image processing for iris recognition applied to the widest population of individuals. This involves the following research:
- Development of a low-cost component to ensure the stabilization and focus of the image;
- Characterization of fine movements of the iris on a representative sample of the population to define a motion model,
- Development of an innovative approach for predicting performance of an iris recognition algorithm based on image quality.
At the end of this project, an experimental system for acquiring high-resolution iris is being developed. It includes a compact adaptive optics system produced with low cost components and coupled to an optimized recognition algorithm. This device is intended to validate the results of studies realized in the project.
Furthermore, it should be emphasized that the development of a high resolution, compact and low cost imaging system, is beyond the scope of the aimed application in this project: iris imaging. It can also create opportunities in other areas of biometrics and even in the long-range surveillance applications.
The consortium is composed of complementary and essential partners in their respective fields. MORPHO will specify requirements, Parrot will provide the necessary components and research centers will focus on developing the algorithm necessary for iris recognition.
Monsieur Serge Meimon (Département d'Optique Théorique et Appliquée) – email@example.com
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
TSP Télécom SudParis
Onera Département d'Optique Théorique et Appliquée
Help of the ANR 459,969 euros
Beginning and duration of the scientific project: January 2013 - 24 Months