Partial fingerprint matching: Fusion of level 2 and level 3 features

Increase in identity frauds, crimes, security there is growing need of fingerprint technology in civilian and law enforcement applications. Partial fingerprints are of great interest which are either found at crime scenes or resulted from improper scanning. These fingerprints are poor in quality and the number of features present depends on size of fingerprint. Due to the lack of features such as core and delta, general fingerprint matching algorithms do not perform well for partial fingerprint matching. By using combination of level1 and level 2 features accuracy of partial matching cannot be increased. Therefore, we utilize extended features in combination with other feature set. Efficacious fusion methods for coalesce of different modality systems perform better for these types of prints. In this paper, we propose a method for partial fingerprint matching using score level fusion of minutiae based radon transform and pores based LBP extraction. To deal with broken ridges and fragmentary information, radon transform is used to get local information around minutiae. Finally, we evaluate the performance by comparing Equal Error Rate (ERR) of proposed method and existing method and proposed method reduces the error rate to 1.84%.