Classification of the Incomplete Fingerprint Image

As the AFIS database more and larger, the processing speeds of AFIS involves a bottleneck. In this case, many biometric researchers have been inspired in studying the fingerprint classification. In this paper, an improve method is introduced which is to aim at solving incomplete fingerprint classification problem in the embedded system. The classification relies on tracking the ridges, which are near the core, by analyzing the block dir image. According to the termination information and the deflexion information of the dermal ridge, each fingerprint will be given one type at least and three types at most by the confidence level. For the five-class problem in the 400 images database the accuracy rate of 99.5% is achieved. Experimental results show the feasibility and validity of the proposed approach in fingerprint classification.