Fingerprint Classication

Every person is believed to have unique fingerprints [4]. This makes fingerprint matching one of the most reliable methods for identifying people [6]. Fingerprint matching is usually done at two di erent levels. At the coarse level, fingerprints can be classified into six main classes: arch, tented arch, right loop, left loop, whorl and twin loop, as shown in Figure 1.The fine-level matching is performed by extracting ridge endings and branching points, called  minutiae, from a fingerprint image (See Figure 2). The similarity between two fingerprints is determined by comparing the two sets of minutiae points. Although the coarse classification does not identify a fingerprint uniquely, it is helpful in determining when two fingerprints do not match. For example, a right loop image should be matched with only other right loop images in the database of fingerprints. When fingerprints from all the ten fingers are available, the coarse level classification of these ten prints drastically reduces the proportion of database images to be matched at the finer level. A human expert can perform coarse-level classification of fingerprints relatively easy. For an automatic system, the problem is much more dicult because the system must take into account the global directions of the ridges as well as their local connectivity to make its decision.

资料全文下载

数据挖掘研究院