Abstract:
Latent fingerprint comparison evidences are used in the court of law for more than 100 years. Manual
matching of latent fingerprints is challenged by subjectivity and inconsistency in terms of results and is not
scalable for large scale applications. Automating the process of latent fingerprint matching will practically
equip forensic examiners in criminal investigation. However, a “lights-out” automated latent fingerprint
matching system is still nascent from being used in a real time environment. Several research challenges
in the development of an automated matching system are identified as: (1) lack of public latent fingerprint
databases available for research, (2) low information content and partial fingerprint availability in latent
fingerprints, (3) presence of background noise and non-linear ridge distortion in latent fingerprints, and (4)
need of an established scientific procedure for matching latent fingerprints. A comprehensive survey in the
growth of latent fingerprint matching, from a computational and algorithms perspective, is provided in this
report. The whole process of automated latent fingerprint matching is divided into five definite stages and
the research gaps in each of the stages are individually analyzed. Also, the limitations in manual matching
of latent fingerprints are studied to gain insights as well as to set a baseline for an automated system. The
major discussions in this survey include: (1) encourage researchers to create and establish results in public
latent fingerprint databases, (2) to focus on the individual stages of a latent fingerprint matching system and
approach them independently, and (3) explore the scope for using some non-standard fingerprint features,
when minutiae extraction becomes challenging.
