Enhancements to latent fingerprints in forensic applications

Latent fingerprint identification is a challenging task in criminal investigation due to the poor quality of ridge impressions and less region of interest on the fingerprint. In this paper, we propose a semi-automated latent fingerprint identification to markup fingerprint landmarks manually using the image enhancement filters which will improve the identification performance in lights-out mode. It uses the global and local adaptive binarization, and global minutia features with ISO/IEC 19794-2 standard fingerprint templates. In the latent fingerprint identification, the fingerprints are matched with rolled fingerprint database from standard law enforcement database. Also, the latent fingerprints are matched with the plain fingerprints database from the data collected using three different live-scanners. The efficacy of the proposed latent fingerprint identification systems is demonstrated on the standard NIST SD-27 (special database-27) latent prints database.

1. Introduction

Fingerprints play an important role in forensic analysis for criminal identification using the clues collected from the crime scene. Eventhough, fingerprint-based identification has been known and used for a very long time [1], it is still challenging task in criminal investigation. The latent prints are the fingerprints which are collected by leaving the finger impressions formed with sweat from fingers. The latent prints have the poor quality of ridge impressions and partial fingerprint area which need ridge enhancements to identify the suspects. As shown in figure 1, there are three different types of finger acquisition, namely, rolled, plain and latent prints. The rolled fingerprints can be acquired by placing the fingerprint on the fingerprint sensor surface and moving it from nail to nail. The plain fingerprints can be captured by simply placing the fingerprint on the sensor surface. The latent fingerprints can be collected from the scene of crime as part of the forensic analysis. The automatic extraction of genuine minutia points from the latent fingerprints becomes difficult due to low fingerprint quality and less area of interest on the fingerprint image.