Fourier cleaning of fingerprint images

摘要

Usage of finger orient imaging for personal identification is one of the oldest and widely used techniques. One of the important steps in fingerprint matching procedure is to reliably extract the discriminating features called minutiae. However, some of the acquired images are of poor quality. Therefore, one must resort to the pre-processing the images. There are various pre-processing techniques has been identified like the Gabor filtering. However, one limiting factor in these e enhancement techniques is their dependence on the extraction of orientation estimates. These estimates remain no longer reliable under some noisy conditions. In this paper we avoided the explicit estimate of the orientation information and relied on the Fourier sinusoidal pattern. We have used Fourier cleaning in the form of notch filter, highpass filter and binarization along with Gabor filter to enhance the fingerprint images. The conducted experiments do suggest a modest raise of around 20–25% in true minutiae matching rate in these poorly visible fingerprint regions.