Abstract
Fingerprint is the most important for the identification of crime scene investigation. Fingerprint technology is not a destroyable one at any time from the early of 20th century. A latent fingerprint is an important main source of the forensic evidences in the court of law. Overlapped finger prints are encountered in latent fingerprints. An automatic match of both latent fingerprints and rolled or plain fingerprints with higher accuracy is important for these applications. A latent impression is of typically of bad quality with detectable background noise which makes the feature extracting and matching of the latent is a notable problem. The problem has been solved by performance gain via feedback from exemplar prints. To solve the problem level-2 and level-3 were proposed. In Level -2 features like minutiae ridge ending, bifurcation is to be introduced. Similarly the level-3 pores like open pores, closed pores fingerprint images also to be introduced.