Resource-efficient latent fingerprint age estimation for Adhoc crime scene forensics: Quality assessment of flat bed scans and statistical features

In the field of biometrics and forensics, age estimation is an important topic, for example when determining the age of human subjects or when assessing the age of a particular taken biometric sample or captured forensic trace from crime scenes. The latter case is investigated in this paper, with the focus on latent fingerprints. Here, the trace age represents the time between deposition and capture of a print, which might assist an investigation in several respects, such as forgery detection (trace freshness inconsistencies) or strengthening a prints evidentiary value in court (linking a suspect to the time of a crime). Non-invasive, high-resolution and very cost-intensive capturing devices have recently been proposed to address this 80 year old challenge of latent print age estimation by studying print time series. This approach is transferred here to a (cheap) off-the-shelf flat bed scanner, enabling a broad variety of researchers to study print aging in the future. Based on 700 time series (including 10% empty substrate series for comparison), a separation performance of 89.2% (eccrine) and 98.7% (sebaceous) was achieved for separating fresh prints and those aged for at least half a day. Results include a comparison to prior used high-resolution Chromatic White Light (CWL) sensors and Confocal Laser Scanning Microscopes (CLSM), also showing the possibility of significant preprocessing simplifications due to decreased capturing distortions as well as the need for further (quantitative) studies.