Conditional Comparison for Binary Fingerprint Images Based on Filtering Technology

摘要

This paper describes a fast and effective approach for fingerprint image preprocessing. It is suitable for the minutiae matching because the approach could filter out error skeletons in the fingerprint image. Traditional preprocessing of fingerprint recognition uses quantization interval to adjust the pixel values in a gray-level fingerprint image to clear the fingerprint ridges. Then, it applies eight direction windows to modify the fault direction of the fingerprint ridges. The third stage is converting the gray-level fingerprint image to black-while one, and thins the fingerprint ridges to their skeletons with one-pixel depth. However, the above procedures may also result error skeletons, such as branches, noises, and gaps. We therefore need to filter those errors out to raise the recognition rate. Experimental results show that our approach is not only simple and fast, but also has the ability to delete all kinds of error skeletons. Hence the approach suggested in the paper should be appropriate for the minutiae matching.