부산대학교 도서관

The aim of this paper is to discuss selected aspects of the emerging trend of digitization in the field of crime scene forensics, known as 'digitized crime scene forensics'. This work summarizes recent findings in the field and discusses the current state of transfer from the analogue to the digital domain. The trace types of latent fingerprints, fibers as well as firearm and lock picking related toolmarks are addressed in respect to the major steps of acquisition, preprocessing, feature extraction and decision making, as support for the subjective expert assessment. Based on the findings, challenges are identified to provide future directions on the issue and to stimulate an increased research in this area. Overall, it can be concluded that a complete digital and automated processing pipeline is missing for most of the considered trace types. The introduction of first digital processing schemes has enabled certain novel, so far not-addressed opportunities, such as separation of overlapped prints, age estimation as well as the application of topographic data and has been able to provide first promising results. But even if parts of this pipeline have been recently automated in first studies, significant challenges remain, such as selecting suitable capturing devices and processing methods, identifying characteristic features and classification strategies as well as employing comprehensive test sets. Most of all, the final classification performance is often not yet good enough to achieve the quality of results obtained in manual investigations, and a significant research effort is required to address the specific needs of numerous trace types and investigation objectives.