Optimizing Parameters for Noise Reduction in Fingerprint Images Using Double Density Dual-Tree DWT

Fingerprint recognition is an important way in person’s identification. Noise are always present in fingerprint images and reducing them the accuracy of minutiae detection is increased. Discreet Wavelet Transform is a method widely used for noise reduction. It is important to find the appropriate mother wavelet and its parameters. Setting the appropriate parameters has impact in getting optimal results. This paper analyzes the performance of Standard DWT, Real 2D dual-tree and complex dual-tree wavelet transform. The parameters analyzed here are the threshold point and the number of decomposition levels. The metric used to measure noise reduction is PSNR, MSR and visual inspection.