Study On Fingerprint Image Enhancement Algorithm Based On NSCT And STFT

Fingerprint image enhancement is one of the important stages in fingerprint identification technology, in order to guarantee the robustness of the fingerprint recognition system, adopting an enhancement algorithm to an original fingerprint image in enhance processing is very necessary.An algorithm based on NSCT and STFT analysis for fingerprint image enhancement is proposed and implemented in this paper. The realization of the algorithm is divided into two stages, the first stage is the fingerprint enhanced pretreatment and the second stage is the filtering enhancement processing. In the first stage, firstly, the original fingerprint image is taken with the NSCT decomposition transformation, then, the high frequency sub-band coefficients of NSCT is taken with hierarchical threshold function to deal with, lastly, the revised coefficients are used in the NSCT inverse transform to reconstruct the fingerprint image. In the second stage, firstly, the fingerprint image is taken with the STFT analysis, then, the local area ridge direction and ridge frequency of fingerprint image is estimated by using the statistical method, lastly, the filter which have corresponding structures with direction and frequency characteristics are constructed to filter the fingerprint image. In the pretreatment stage, our algorithm can get the highest PSNR compared with other denoising algorithms, its PSNR is nearly3.2dB higher than the that of algorithm which based on wavelet denoising and nearly1.5dB higher than the that of algorithm which based on Contourlet denoising. In the second stage, our algorithm feature extraction’s average error rate is nearly10.69%less than the Gabor enhancement method. Experimental results demonstrate that, the algorithm based on NSCT and STFT analysis for fingerprint image enhancement can reduce the noise effectively and have better retention of image edge texture the detail information, besides, the algorithm not only can make the fingerprint lines clearly but also have very good effects in the fingerprint lines fault and adhesion place.