On using soft biometrics in forensic investigation

Abstract

This chapter addresses the usage of biometric recognition tools in the context of forensic investigations. In particular, the authors are concerned with the extraction of evidence from video sequences captured by surveillance cameras. In such scenarios many of the biometric traits traditionally used for recognition purposes, such as fingerprints, palmprints or iris, are not available. Therefore, the focus is on the extraction of soft biometrics, which encompasses personal characteristics used by humans to recognize or help to recognize an individual. This work starts by reviewing how forensic casework relying on surveillance video information is conducted nowadays. Then, a software platform, BioFoV, is proposed to automate many of the required procedures and including some initial implementation of soft biometric extraction tools. Furthermore, novel biometric methods to analyze human gait and facial traits are described and experimentally validated as a demonstration of future perspectives in soft biometrics.