A Computational Approach to Edge Detection
IEEE Transactions on Pattern Analysis and Machine Intelligence
BioID'11 Proceedings of the COST 2101 European conference on Biometrics and ID management
| Hi-index | 0.00 |
Malicious fingerprint traces at crime scenes might be an upcoming challenge for forensic investigations and should be differentiated from valid traces to avoid false accusations. Schwarz [8] introduces an approach for printing amino acids using ink-jet printers. This technique, originally intended for quality assessment might be misused to produce such malicious traces. Hence, Kiltz et al. [5] perform a subjective assessment of traces to find potential detection properties. Our scope in this paper is to find and evaluate algorithms for the feature extraction of malicious fingerprint traces to be able to detect such traces automatically. We use the HoughCircles algorithm, which is capable of detecting circular dots on a particular trace. In our experiments we perform an analysis of the extracted features using statistical approaches to evaluate their suitability for an automatic classification. Our test setup includes 60 fingerprint samples on overhead foil, which originate from two different printers and one human being for comparison.