Evaluation of data treatment techniques for improved analysis of fingerprint images

We evaluated four different methods for data treatment of fingerprint images. The first two methods (Lee's method, and Iterative Automated Noise Filtering method) are used to remove signal-independent additive noise from digital images of fingerprints. The two other methods (low-pass and high- pass wavelet filtering), based on the wavelet transform, are employed not only to reduce the noise, but also to enhance the information of the fingerprints, especially details at the ridges. To evaluate these four filtering techniques we analyzed images of various fingerprints provided by the Police Department of Knoxville, Tennessee, and recorded digitally in the laboratory using inexpensive apparatus. Experimental results show that the methods are effective in diverse cases and have potential in forensic analysis of fingerprints.

Duke Scholars

Author Tuan Vo-Dinh Biomedical Engineering