On high resolution palmprint matching

This paper proposes a high resolution palmprint biometric recognition system that matches incomplete or degraded palmprints. Most of the existing work in this field was developed for commercial applications, usually for low resolution images. For high resolution palmprints, such as the ones used by forensic experts, the same techniques tend not to explore the full potential of the available biometric information. When dealing with latent palmprints, it is very common to find degradations, which can often be modeled by added Gaussian or salt & pepper noise, as well as motion blur due to the sliding of the hand. A system able to handle these degradations is presented, including a new technique for motion blur detection and compensation, based on a technique initially developed for low resolution barcode reading. Promising results are obtained.