Study On Automatic Segmentation And Recognition Of Slap Fingerprint

With the popularity of fingerprint technology, especially the application on iPhone mobile and 3rd identity card(ID) of PRC, fingerprint technology expands with high speed beyond the previous, and is gradually accepted by common people. The emergence of fake fingerprint does throw doubt on their choices. To improve the safety and reliability of fingerprint recognition, slap fingerprint is born at the right moment. For it provides a convenient and shortcut way to capture multi-finger images, the slap fingerprint is not only being applied in police field originally but also being expanded to civil field now.First, an automatic segmentation algorithm for slap fingerprint is proposed, which is based on frequency domain analysis. The main process of this algorithm includes: foreground and background separating, noise detection and removing, crease detection, the removing of non-fingertip domain, border frame calibration. Comparing to the traditional slap segmentation algorithm, this algorithm is with higher accuracy and faster speed.Second, a new segmentation and matching method for the slap fingerprint is proposed. Minutiae are extracted from the original slap directly, and then they are divided into four sets according to the statistical histogram. To adapt to the possible zooming of different slap images captured from HD mobile phones, RST(Rotation, Scaling, and Translation) invariance features are used to match two sets, which can improve the compatibility of slap fingerprints from different scanning modes.Because the slap fingerprint can provide higher reliability than the single fingerprint, it is applied in the artwork investment management. We have developed an MFP(Multi-Finger-Pass) system, which has been tested for 3 months and the testing result shows that the proposed algorithm can meet the practical requirement.