A Comparison between Backpropagation Neural Network and Seven Moments for More Accurate Fingerprint Video Frames Recognition

Abstract

In this modern age of electronic interactions, more secure methods are required to protect vital information. Passwords are indeed a prominent and secure method, but they are subject to being forgotten, especially if they are long and complex. A more efficient way is the use of human fingerprints, which are unique to each person. No two people would have the same fingerprint even if they were a twin, which makes it a very secure method that cannot be duplicated or forgotten. This research aims to compare seven moments and backpropagation for more accurate fingerprint recognition within video frames. The first method is the "seven moments," and the second method is the Backpropagation Neural Network (BPNN), both applied to the interest points that are extracted from each frame. For extracting the interest points from each one of the frames, Smallest Univalue Segment Assimilating Nucleus (SUSAN), a corner detector, was employed. Multiple examples of video frames were used in comparison, and the findings demonstrated that the BPNN approach was more accurate even when the fingerprint had a significant amount of corrupted data or unclear image pixels.

Keywords

• BackPropagation, Fingerprint, Recognition, Seven Moments, SUSAN Detector.