A Fingerprint Image Denoising And Enhancement Method Based On Complex Wavelet

The performance of fingerprint identification system is mostly limited by the quality of fingerprint images. So it is necessary to preprocess the fingerprint images. Because the print ink of fingerprint image is uneven, the correct estimation ridge direction of origin image can not be sure. An image denoising and enhancement method based on complex wavelet transform is proposed in this dissertation, which is independent of fingerprint directional diagram.The improved OTSU algorithm and genetic algorithm are applied to fingerprint image segmentation. Compared with OTSU algorithm and PCNN algorithm, results of experiment show fingerprint details are reserved significantly and the good image segmentation effect after background separated is achieved. Donoho threshold,Bayes threshold,GCV threshold,SURE threshold,soft_hard threshold,context model,hybrid model and bivariate shrinkage function are applied to denoise experiment data in DWT,DTRWT and DTCWT domain respectively . The experiment results are shown that the effect of the image denoising method using bivariate shrinkage technique performs the other methods obviously in our experiments.At last, an improved bivariate shrinkage denoising method based on crossscale and interscale coefficients constraints is presented, which is aiming at optimizing contrast of image features while minimizing image noise. With the advantage of the directivity and shift-invariant of complex wavelet transform, the subtle features are retained while noise is efficiently suppressed. Compared with the image denoising method using bivariate shrinkage function, the experiment result shows the PSNR is enhanced 0.3~ 0.5dB using this method. An adaptive function for image enhancement ,which can emphasize the low contrast features and protect the strong contrast features from distortion ,is given in this paper. The experimental result indicates that the quality of enhanced image is improved obviously.