Quality Estimation for Vascular Pattern Recognition

The quality of captured samples is a critical aspect in biometric systems. In this paper we present a quality estimation algorithm for vascular images, which uses global and local features based on a Grey Level Co-Occurrence Matrix (GLCM) and optionally available metadata. An evaluation of the algorithm using different processing methods and vein sample databases shows convincing results: disregarding low estimated quality sample images helps to increase the performance. Moreover, metadata gives accurate indications on sample quality. The algorithm works on low level raw images, it is fast and therefore qualified to be used in feedback mode during enrolment or verification operation.