The BioSecure Benchmarking Methodology for Biometric Performance Evaluation

Abstract

Measuring real progress achieved with new research methods and pinpointing the unsolved problems is only possible within a well defined evaluation methodology. This point is even more crucial in the field of biometrics, where development and evaluation of new biometric techniques are challenging research areas. Such an evaluation methodology is developed and put in practice in the European Network of Excellence (NoE) BioSecure. Its key elements are: open-source software, publicly available biometric databases, well defined evaluation protocols, and additional information (such as How-to documents) that allow the reproducibility of the proposed benchmarking experiments. As of this writing, such a framework is available for eight biometric modalities: iris, fingerprint, online handwritten signature, hand geometry, speech, 2D and 3D face, and talking faces. In this chapter we first present the motivations that lead us to the proposed evaluation methodology. A brief description of the proposed evaluation tools follows. The multiple possibilities of how this evaluation methodology can be used are also described, and introduce the other chapters of this book that illustrate how the proposed benchmarking methodology can be put into practice.