Iris biometrics security, performance enhancement patent issued to Saudi Arabian school
King Saud University has been awarded a patent for an architecture of multi-resolution iris recognition software intended to provide better security and performance than existing systems.
The system is claimed to use a “multi-algorithm, multi-biometric, and multi-instance” approach to processing data extracted from the periocular region around the iris, as well as the iris itself.
The performance of the deep learning-based iris matching is improved by the software being able to work around multiple “challenging issues of unconstrained imaging environments” through “advanced feature representation techniques such as multi-resolution image analysis, fine-tuning, transfer learning” and the use of multiple algorithms, biometric modalities (iris and periocular and instances, according to U.S. patent 11,762,969.
The patent also covers the use of cancelable biometric templates and cryptographic data protection. It’s all cancelable remotely in case of a breach, a concept which has been showing up repeatedly in recent biometrics research. In that case a template can be changed for future use.
The University says in an announcement that the technology described can be used in border control, healthcare, banking and other applications.