FPGA implementation of CMF for embedded palm biometric system

Most multimodal palm biometric systems are still PC-based, because of the complicated image processing algorithms necessary to process biometric data. In this paper we demonstrate how to implement Complex Matched Filter in FPGA based systems for real-time palmprint and palm vein image processing. CMF approach is based on matched filtering with rotated line extraction kernels, however, CMF requires less computational resources and it obtains additional angular information about the extracted biometric features. This information is valuable in feature description and recognition process. Topics that are covered in this paper include optimization of used multipliers and real-time data processing without use of external RAM resources.