A novel wavelet based thresholding for denoising fingerprint image

The robustness of a fingerprint authentication system depends on the quality of the fingerprint image. Denoising of the fingerprint image is indispensable to get a noise free image. In this paper, a novel method is proposed to remove Gaussian noise present in fingerprint image using Stationary Wavelet Transform (SWT), a threshold based on Golden Ratio and weighted median. First decompose the input image using SWT and apply the new modified universal threshold to the wavelet coefficients using hard and soft thresholding. Then apply Inverse Stationary Wavelet Transform (ISWT) to get the noise free image. The different kinds of wavelet filters such as db1, db2, db4, sym2, sym4, coif2 and coif4 for different noise levels are performed, among which db2 outperformed. In this study, experiments have been conducted on the fingerprint database FVC2002. The Peak Signal-to-Noise Ratio (PSNR), Signal-to-Noise Ratio (SNR), Root Mean Square Error (RMSE) and Mean Square Error (MSE) of the new modified universal threshold combined with hard thresholding is improved compared with the existing methods.