Speed-optimized fingerprint image enhancement for embedded systems

Fingerprint recognition is the most widely used form of biometric identification. Many high performance recognition algorithms exists, which can reliably identify even poor fingerprint images, but most of these algorithms require great processing power implying the use of some computer systems to do the actual recognition. However, more and more applications require fingerprint recognition to be done in small embedded system with low processing power but within an acceptable timeframe. This paper focuses on some possible speed-up techniques for image enhancement and their effect on recognition performance.