Research

In the DMCS Biometric Laboratory researches concerning multimodal biometric authentication systems are conducted. The work is mainly focused on iris and face recognition. Additionally solutions for authentication based on fingerprint and other modalities are developed. One of the main streams of the researches is analysis of 4D signals (3D range images over time) of face and hand biometric features. Moreover, the team is involved in practical implementations of the developed algorithms and methods. For this purpose many issues related to the specific video data acquisition must be solved including hardware platforms for its real-time processing.
One of the remaining problems in biometric recognition is the implementation of its efficient hardware systems. The difficulty arises from the fact that the methods forming part of this type of authentication process are fairly complex. Furthermore, next generation algorithms for biometric recognition will become even more complex to enhance the reliability and functionality of currently used solutions. An example of such an intensive processing task is iris image quality assessment and segmentation using video signal obtained at-a-distance and on-the-move. Thus, processing robustness and predictable analysis time in such dedicated architectures may be particularly important. The lab research interests include the field-programmable gate array and digital signal processor hardware systems for advanced image processing with an emphasis on the biometric applications, namely the study of iris- and face-based authentication.