Maik Schott

Unweighted fusion in microphone forensics using a decision tree and linear logistic regression models

For the exemplarily chosen domain of microphone forensics we show that media forensics can strongly benefit from combining statistical pattern recognition (using supervised classification) and unweighted information fusion (on the example of match-, rank- and decision level fusion). The practical results presented show that, by using a carefully selected fusion strategy and two multi-class classifiers (a decision tree and linear logistic regression models), the accuracy achieved in practical testing can be increased to 100%. This result is based on first tests on two sets of four and seven different microphones. For each of those microphones ten reference samples are recorded in ten different locations and are used in the ratio 80% to 20% for supervised training and testing by the two classifiers. The overall positive tendency indicates that microphone forensics might become an important security mechanism for the verification of source authenticity. Recent gunshot classification approaches, which try to determine the gun used in gunshot audio recordings, have the problem that they rely on carefully controlled conditions, amongst them the fact that the microphone used for all evaluations has to remain the same. A microphone classification approach as introduced here would allow for similarity estimation for microphones and thereby would enable exchanging microphones in such a gunshot classification approach without complete loss of confidence. Furthermore microphone forensics could be used in provenance verification of digital audio media to verify the microphone used for recordings to be submitted into secure long term archiving systems.

A Context Model for Microphone Forensics and its Application in Evaluations

In this paper we first design a suitable context model for microphone recordings, formalising and describing the involved signal processing pipeline and the corresponding influence factors. As a second contribution we apply the context model to devise empirical investigations about: a) the identification of suitable classification algorithms for statistical pattern recognition based microphone forensics, evaluating 74 supervised classification techniques and 8 clusterers; b) the determination of suitable features for the pattern recognition (with very good results for second order derivative MFCC based features), showing that a reduction to the 20 best features has no negative influence to the classification accuracy, but increases the processing speed by factor 30; c) the determination of the influence of changes in the microphone orientation and mounting on the classification performance, showing that the first has no detectable influence, while the latter shows a strong impact under certain circumstances; d) the performance achieved in using the statistical pattern recognition based microphone forensics approach for the detection of audio signal compositions.

Modelling watermark communication protocols using the CASPER modelling language

In cryptography it is common to evaluate the security of cryptographic primitives and protocols in a computational model, with an attacker trying to break the primitive or protocol in question. To do so formalisation languages like CASPER or CSP (Communication Sequential Processes) and model checkers like FDR (Failures-Divergences Refinement) are used for automatic or semi-automatic machine-based security verification. Here we transfer the idea of machine-based verification of the security of communication protocols from cryptography to the domain of digital watermarking based media security protocols. To allow for such a mainly automatic verification approach, we introduce and illustrate in this paper a six step procedure for the modelling and verification of watermark communication protocols based on application scenario descriptions. The six steps are: First, a modelling of the used communication network and application scenario (as a task) in XML-structures, second, a path search comparing the network and the task and identifying possible watermarking channels, third, a path selection selecting one watermarking channel from the identified alternatives for the protocol realisation, fourth, an automatic CASPER protocol generation from the selected alternative followed by manual adjustments (if necessary), fifth, the CASPER compilation into CSP and sixth, the protocol security(confidentiality, integrity and authenticity) verification via the FDR model checker.

Context-based approach of separating contactless captured high-resolution overlapped latent fingerprints

Overlapped latent fingerprints occurring at crime scenes challenge forensic investigations, as they cannot be properly processed unless separated. Addressing this, Chen et al. proposed a relaxation-labelling-based approach on simulated samples, improved by Feng et al. for conventionally developed latent ones. As the development of advanced contactless nanometre-range sensing technology keeps broadening the vision of forensics, the authors use a chromatic white light sensor for contactless non-invasive acquisition. This preserves the fingerprints for further investigations and enhances existing separation techniques. Motivated by the trend in dactyloscopy that investigations now not only aim at identifications but also retrieving further context of the fingerprints (e.g. chemical composition, age), a context-based separation approach is suggested for high-resolution samples of overlapped latent fingerprints. The authors conception of context-aware data processing is introduced to analyse the context in this forensic scenario, yielding an enhanced separation algorithm with optimised parameters. Two test sets are generated for evaluation, one consisting of 60 authentic overlapped fingerprints on three substrates and the other of 100 conventionally developed latent samples from the work of Feng et al. An equal error rate of 5.7% is achieved on the first test set, which shows improvement over their previous work, and 17.9% on the second.

Sequence detection of overlapping latent fingerprints using a short-term aging feature

This paper presents a novel approach for sequence detection of overlapping latent fingerprints for crime scene forensics. The approach involves a contactless and nondestructive acquisition of untreated fingerprints by using a Chromatic White Light (CWL) sensor. Based on 40 time series with 1160 samples, our approach shows how an aging feature called Binary Pixel for measuring the short-term decay of fingerprint samples can be used in combination with overlapping fingerprint separation methods for sequence detection. This allows differentiating which latent fingerprint was placed first and which later and thus provides contextual information for criminal investigations. Our experiments show promising results with a detection accuracy of at least 70%, regardless of the initial age of both the older and newer fingerprint.

Separation of high-resolution samples of overlapping latent fingerprints using relaxation labeling

The analysis of latent fingerprint patterns generally requires clearly recognizable friction ridge patterns. Currently, overlapping latent fingerprints pose a major problem for traditional crime scene investigation. This is due to the fact that these fingerprints usually have very similar optical properties. Consequently, the distinction of two or more overlapping fingerprints from each other is not trivially possible. While it is possible to employ chemical imaging to separate overlapping fingerprints, the corresponding methods require sophisticated fingerprint acquisition methods and are not compatible with conventional forensic fingerprint data. A separation technique that is purely based on the local orientation of the ridge patterns of overlapping fingerprints is proposed by Chen et al. and quantitatively evaluated using off-the-shelf fingerprint matching software with mostly artificially composed overlapping fingerprint samples, which is motivated by the scarce availability of authentic test samples. The work described in this paper adapts the approach presented by Chen et al. for its application on authentic high resolution fingerprint samples acquired by a contactless measurement device based on a Chromatic White Light (CWL) sensor. An evaluation of the work is also given, with the analysis of all adapted parameters. Additionally, the separability requirement proposed by Chen et al. is also evaluated for practical feasibility. Our results show promising tendencies for the application of this approach on high-resolution data, yet the separability requirement still poses a further challenge.

Car-seat occupancy detection using a monocular 360° NIR camera and advanced template matching

The integration of seat occupancy detection systems is one of the most recent developments in automobile production. These systems prevent the deployment of airbags at unoccupied seats, thus avoiding the considerable cost imposed by the replacement of airbags. Seat-occupancy detection system can also be used to improve passenger comfort, e.g. by an occupation-dependent control of air-conditioning systems. This paper describes an inexpensive and versatile optical seat-occupancy detection system. Different approaches to pattern matching and the impact of local normalization, edge detection, multi-algorithm and temporal matching-score fusion are evaluated for each individual seat using a test set of 53,928 frames further classified in uniform and non-uniform illumination conditions. The results of these tests yield Equal Error Rates for uniform/non-uniform illumination of as low as 3.05%/1.68% for the front left seat, 2.17%/0.69% for the front right seat, 5.86%/4.01% for the rear left seat, 10.99%/11.07% for the rear center seat and 5.63%/1.84% for the rear right seat. The test results indicate that at least the two seat rows should be treated differently in terms of the selection of classification algorithms.

Image annotation watermarking: nested object embedding using hypergraph model

In this paper, we introduce to the special domain of image annotation watermarking, based on embedding of hierarchical data related to objects into user-selected areas on an image. In comparison to earlier methods, the main goal of the work presented here is to provide a specific robustness, specifically against cropping, in a way that preserves hierarchical object relations even after retrieval from a partial image, which has been cut from the original (pure security aspects are of limited relevance for our application). We identify two initial categories for such relations, visual-functional and visual-spatial and suggest a novel coding scheme for the first of these. Our prototypical system consists of a new ontology-based interactive editor and a watermarking scheme, which is an extension of previously suggested block-based image watermarking towards the specific requirements for annotation watermarking. Our initial experiments of the new approach include evaluation of appropriate synchronization patterns and issues of their threshold-based exhaustive search. Further, we present results from robustness tests, which are based on cropping and lossy JPEG compression. Our findings show that the suggested method is capable to restore payload in a hierarchy-preserving way after up to 50% JPEG compression, with a low to medium loss of transparency, which has been evaluated subjectively. Further, we observe that even when limiting exhaustive search to those watermark candidates showing the highest embedding energy, we are able to correctly detect between 55% and 100% of cropped objects.

Legally compatible design of digital dactyloscopy in future surveillance scenarios

Innovation in multimedia systems impacts on our society. For example surveillance camera systems combine video and audio information. Currently a new sensor for capturing fingerprint traces is being researched. It combines greyscale images to determine the intensity of the image signal, on one hand, and topographic information to determine fingerprint texture on a variety of surface materials, on the other. This research proposes new application areas which will be analyzed from a technical-legal view point. It assesses how technology design can promote legal criteria of German and European privacy and data protection. For this we focus on one technology goal as an example.

Extending the Clark-Wilson Security Model for Digital Long-Term Preservation Use-cases

A continuously growing amount of information of today exists not only in digital form but were actually born-digital. These informations need be preserved as they are part of our cultural and scientific heritage or because of legal requirements. As many of these information are born-digital they have no analog origin, and cannot be preserved by traditional means without losing their original representation. Thus digital long-term preservation becomes continuously important and is tackled by several international and national projects like the US National Digital Information Infrastructure and Preservation Program [1], the German NESTOR project [2] and the EU FP7 SHAMAN Integrated Project [3]. In digital long-term preservation the integrity and authenticity of the preserved information is of great importance and a challenging task considering the requirement to enforce both security aspects over a long time often assumed to be at least 100 years. Therefore in a previous work [4] we showed the general feasibility of the Clark-Wilson security model [5] for digital long-term preservation in combination with a syntactic and semantic verification approach [6] to tackle these issues. In this work we do a more detailed investigation and show exemplarily the influence of the application of such a security model on the use cases and roles of a digital long-term preservation environment. Our goals is a scalable security model - i.e. no fixed limitations of usable operations, users and objects - for mainly preserving integrity of objects but also ensuring authenticity.