Indexing iris images using the Burrows-Wheeler Transform

In most biometric identification systems, the input biometric data has to be compared against that of every identity in the database in order to determine the identity of the input. A major problem with this approach is the impact on response time which can increase significantly with the size of the database. In certain applications such as real time monitoring, this delay may not be acceptable. In this work, we propose a method for indexing iris images for rapid identity retrieval. Every entry in the database is assigned an index code based on which a small subset is retrieved and matched in response to a query. The basis of our approach is the sorted context property of the Burrows Wheeler Transform, a popular transformation used in data compression. Experiments on the CASIA version 3 iris database show a significant reduction in both search time and search space.