Refine
Year of publication
- 2011 (2) (remove)
Document Type
Language
- English (2)
Has Fulltext
- no (2)
Is part of the Bibliography
- yes (2)
Keywords
Institute
Technologies for advancing and supporting criminalistic forensic are an upcoming challenge with rising importance within the domain of multimedia security and forensic. For example the acquisition and automated analysis of latent fingerprints, using high resolution 3D scanners, appears to be a promising area of research. Within the paper we will give a detailed demarcation of biometric and forensic fingerprint analysis. We will introduce the aim of a partly automated process for forensic fingerprint acquisition, detection, and processing. Based on the idea of splitting the overall scan process into a fast, low resolution coarse scan and a high resolution detailed scan, different approaches and algorithms has to be evaluated, whether they are feasible for detecting latent fingerprints. Referring to the subject of biometrics and emerging applications, this work will show an approach of a novel Fourier-based spectral texture feature for detecting latent fingerprints in low resolution grey-scale images. Our work is based on data that is acquired using a chromatic white light sensor (CWL), which provides intensity and topographic information of the scanned surface areas. While using a very limited set of different surfaces so far, our first experiments have shown good results on flat and non structured surfaces. Using the presented feature, it is possible to detect latent fingerprints on these surfaces. Even on slightly structured surfaces, like wood imitation, the application of the spectral density feature yields promising results. However the first evaluation of the spectral density feature has also shown some serious limitations of its use on low resolution images.
The automated forensic analysis of latent fingerprints poses a new challenge. While for the pattern recognition aspects involved, the required processing steps can be related to fingerprint biometrics, the common biometric model needs to be extended to face the variety of characteristics of different surfaces and image qualities and to keep the chain of custody. Therefore, we introduce a framework for automated forensic analysis of latent fingerprints. The framework consists of a generic process model for multi-branched process graphs w.r.t. security aspects like integrity, authenticity and confidentiality. It specifies a meta-model to store all necessary data and operations in the process, while keeping the chain of custody. In addition, a concept for a technical implementation of the meta-model is given, to build a container format, which suits the needs of an automated forensic analysis in research and application.