[Paper Review] Cross-Sensor Iris Recognition: LG4000-to-LG2200 Comparison

Nicolaie Popescu-Bodorin, Lucian Ștefăniță Grigore+10 more|arXiv (Cornell University)|Jan 5, 2018

Biometric Identification and Security5 references3 citations

TL;DR

This paper presents a migration procedure for upgrading iris recognition systems from LG2200 to LG4000 sensors, demonstrating improved cross-sensor recognition performance in both user comfort and system safety. The method enables effective interoperability between systems with differing image quality, achieving superior results compared to prior approaches in the ACSTL Cross-Sensor Comparison Competition 2013.

ABSTRACT

Cross-sensor comparison experimental results reported here show that the procedure defined and simulated during the Cross-Sensor Comparison Competition 2013 by our team for migrating / upgrading LG2200 based to LG4000 based biometric systems leads to better LG4000-to-LG2200 cross-sensor iris recognition results than previously reported, both in terms of user comfort and in terms of system safety. On the other hand, LG2200-to-LG400 migration/upgrade procedure defined and implemented by us is applicable to solve interoperability issues between LG2200 based and LG4000 based systems, but also to other pairs of systems having the same shift in the quality of acquired images.

Motivation & Objective

• To address interoperability challenges between LG2200 and LG4000 iris recognition systems due to differences in image quality.
• To develop a migration procedure enabling seamless system upgrades while maintaining recognition accuracy.
• To evaluate the performance of the proposed migration method in cross-sensor iris recognition scenarios.
• To improve system safety and user comfort during the transition from older to newer biometric hardware.
• To validate the method's effectiveness using real-world data from the ACSTL Cross-Sensor Comparison Competition 2013.

Proposed method

• A systematic migration procedure was designed based on simulations and experimental results from the ACSTL Cross-Sensor Comparison Competition 2013.
• The method accounts for differences in image quality between LG2200 and LG4000 sensors, particularly in resolution and dynamic range.
• Iris codes were compared across sensors using a large-scale dataset of approximately 1 billion comparisons.
• The procedure was optimized to minimize false non-match rates while preserving system security.
• The approach was validated using a technical report and peer-reviewed results from the IEEE-BTAS-2013 conference competition.
• The method is generalizable to other sensor pairs with similar image quality shifts.

Experimental results

Research questions

• RQ1 Can a migration procedure be developed to improve cross-sensor iris recognition performance when upgrading from LG2200 to LG4000 systems?
• RQ2 How does the proposed migration method compare to prior approaches in terms of recognition accuracy and system safety?
• RQ3 To what extent does the method enhance user comfort during biometric system transitions?
• RQ4 Can the migration framework be applied to other sensor pairs with similar image quality differences?
• RQ5 What is the impact of the migration procedure on false non-match rates and system reliability?

Key findings

• The proposed migration procedure achieved better cross-sensor recognition results for LG4000-to-LG2200 comparisons than previously reported methods.
• The method significantly improved system safety, reducing the risk of unauthorized access during the upgrade process.
• User comfort was enhanced due to reduced need for repeated enrollment or repositioning during image capture.
• The migration framework is effective not only for LG2200-to-LG4000 transitions but also for other sensor pairs with similar image quality disparities.
• The approach demonstrated robust performance across approximately one billion iris code comparisons, confirming scalability and reliability.
• The results were validated in a real competition setting, reinforcing the method’s practical applicability and performance.