Efficient Fingerprint Image Enhancement for Mobile Embedded Systems

Fingerprint image enhancement is an important step in a fingerprint verification system. The enhancement process, however, are often not applied to mobile embedded devices in which floating-point processing units (FPU) are absent. Earlier Hong and Jain reported a fingerprint enhancement algorithm based on the Gabor Filter. This algorithm and its derivatives have been proved to be quite effective in improving the fingerprint verification reliability. In this paper, we present an efficient implementation of this algorithm in an embedded system environment. In our implementation, fixed-point arithmetic is used to replace the floating-point operations. Moreover, a special Gabor filter parameter selection constraint is also proposed to reduce the computing complexity of the kernel generation step. Experimental results show that our new approach achieves significant speed improvement and is almost as effective as the traditional floating-point based implementation.