Showing papers by "Jana Dittmann published in 2009"

Microphone Classification Using Fourier Coefficients

Abstract: Media forensics tries to determine the originating device of a signal. We apply this paradigm to microphone forensics, determining the microphone model used to record a given audio sample. Our approach is to extract a Fourier coefficient histogram of near-silence segments of the recording as the feature vector and to use machine learning techniques for the classification. Our test goals are to determine whether attempting microphone forensics is indeed a sensible approach and which one of the six different classification techniques tested is the most suitable one for that task. The experimental results, achieved using two different FFT window sizes (256 and 2048 frequency coefficients) and nine different thresholds for near-silence detection, show a high accuracy of up to 93.5% correct classifications for the case of 2048 frequency coefficients in a test set of seven microphones classified with linear logistic regression models. This positive tendency motivates further experiments with larger test sets and further studies for microphone identification.

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

Abstract: 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.

Automotive IT-Security as a Challenge: Basic Attacks from the Black Box Perspective on the Example of Privacy Threats

Abstract: Since automotive IT is becoming more and more powerful, the IT-security in this domain is an evolving area of research. In this paper we focus on the relevance of the black box perspective in the context of threat analyses for automotive IT systems and discuss typical starting points and implications of respective attacks. We put a special focus on potential privacy issues, which we expect to be of increasing relevance in future automotive systems. To motivate appropriate provision for privacy protection in future cars we discuss potential scenarios of privacy violations. To underline the relevance even today, we further present a novel attack on a recent gateway ECU enabling an attacker to sniff arbitrary internal communication even beyond subnetwork borders.

On the need of data management in automotive systems

Abstract: In the last decade, automotive systems changed from traditional mechanical or mechatronical systems towards software intensive systems, because more and more functionality has been implemented by software. Currently, this trend is still ongoing. Due to this increased use of software, more and more data accumulates and thus, has to be handled. Since it was no subject up to now to manage this data with software separatly, we think that it is indispensable to establish a data management system in automotive systems. In this paper we point out the necessity of data management, supported by exemplary scenarios, in order to overcome disadvantages of current solutions. Further, we discuss main aspects of data management in automotive systems and how it could be realized with respect to the very special restrictions and requirements within such a system.

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

Abstract: 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.

A new forensic model and its application to the collection, extraction and long term storage of screen content off a memory dump

Abstract: In this paper we show how to extract graphics content within a memory dump of a Windows-based system. This includes the assurance of integrity and authenticity of evidence gathered this way using cryptographic mechanisms. We introduce a forensic data model and investigate different forensic analysis steps within a phase-oriented manner to classify potential forensic methods. Furthermore we discuss approaches for long term preservation for the forensic data aquired from the memory dumps to ensure authenticity and integrity.

The impact of information fusion in steganalysis on the example of audio steganalysis

Abstract: Information fusion tries to determine the best set of experts in a given problem domain and devise an appropriate function that can optimally combine the decisions of the individual experts. Only few systematic approaches to information fusion exist so far in the signal processing field of steganalysis. Under the basic assumption that steganalysis can be seen as a statistical pattern recognition process like biometrics, a state of the art five level information fusion model known from biometrics is transferred to steganalysis as well as statistical detectability evaluations for watermarking algorithms and its applicability is evaluated in practical testing. The primary test goal for these evaluations is to measure the impact of fusion on the classification accuracy. Therefore a match and decision level fusion are performed here for three selected data hiding algorithms (one steganography and two watermarking), two feature extractors and five different classifiers. For the test heterogeneous audio test sets are used for content independent training and testing. The secondary test goal of this work is to consider the impact of the key selection assumption on the accuracy of the classification in steganalysis. The results show for the test cases an increase of the classification accuracy for two of the three tested algorithms by match level fusions, no gain by decision level fusion and a considerably small impact of the key selection assumption on the statistical detectability.

From the Computer Incident Taxonomy to a Computer Forensic Examination Taxonomy

Abstract: Forensic investigations are usually conducted to solve crimes committed using IT-systems as pertetrator and/or victim. However, depending on the size of IT-system, also non-malicious incidents can be investigated using the same, methodological and proven techniques. Based upon the principles contained in the well-known Computer Incident Taxonomy [1], this paper proposes the establishment a common language for the description of computer forensic examinations, both in malicious and non-malicious incidents. Additionally this taxonomy helps performing a forensic examination in establishing answers to a set of well-definied questions during such an examination. The usefulness of the proposed Forensic Examination Taxonomy is shown using a malicious and a non-malicious example.

Vertrauenswürdige und abgesicherte Langzeitarchivierung multimedialer Inhalte

Abstract: Inhaltsverzeichnis Inhaltsverzeichnis 1 Einleitung, Motivation und Überblick 1

AnnoWaNO: An annotation watermarking framework

Abstract: In this paper we visit annotation watermarking as a tool to embed information about a cover or parts thereof in the cover. We describe a framework for image annotation watermarking, especially Nested Object Annotation Watermarking. This was introduced in earlier works to embed not only textual and/or semantic descriptions as annotations but also their relationships with each other.

Biometric hash generation and user authentication based on handwriting using secure sketches

Abstract: Contrary to cryptographic hash functions, which calculate the same value for only the same input, biometric hash functions have to generate identical hashes from varying input data of the same person. In this paper a secure sketch method is suggested that was developed to generate hash values from dynamic handwritten data. The presented secure sketch method is motivated by the framework suggested by Dodis et al. ([1]) and the face based secure sketch algorithm of Sutcu et al. ([2]). The results of the proposed method are compared to those of another biometric hashing algorithm from recent work of the authors as reference algorithm. On one side the evaluation results showing a very promising performance of the secure sketch method in case of reproducibility (best result 96%) and collision probability (best result 3%). Compared to the reference algorithm this results corresponds to an improvement of approx. 20% for reproducibility and approx. 45% for collision probability based on the same test setup. On the other side the verification performance of the reference algorithm, measured by EER, is higher than for the secure sketch method. While the reference algorithm determines an EER of 0.0233 in the best case, the secure sketch method calculates an EER of 0.0559.

Security-relevant challenges of selected systems for multi-user interaction

Abstract: One important goal in the field of multi-user interaction is to support collaborative work of several users as ergonomic as possible. Unfortunately, security-relevant aspects were neglected in the past. Therefore, we study in this contribution the risks and challenges for security of such collaborative working environments on the basis of five selected pen and gesture-based input techniques. We show that the underlying technologies (Anoto pens, Wii Remotes, DiamondTouch, FTIR Table tops, Microsoft Surface) do have deficits, in particular regarding the insurance of user authenticity and data integrity, and that collaborative working brings new challenges for formal security models. We discuss some of the major challenges on situation and context recognition for dynamic role assignment based on a scenario from the field of energy engineering and point out that several of the underlying problems are of special importance for the development of reliable collaborative multimedia applications for object organization and exchange.

Multi-biometric fusion for driver authentication on the example of speech and face

Abstract: Nowadays biometrics becomes an important field in IT security, safety and comfort research for automotive. Aims are automatic driver authentication or recognition of spoken commands. In this paper an experimental evaluation of a system is presented which uses a fusion of three biometric modalities to verify the authorized drivers out of a limited group of potential persons such as a family or small company which is a common use case for automotive domain. The goal is to show the tendency of biometric verification performance in such a scenario. Therefore a multi-biometric fusion is carried out based on biometric modalities face and voice in combination with the body weight. The fusion of the three modalities results in a relative improvement of 140% compared to the best individual result with regard to the used measure, the equal error rate.

Nested object watermarking

Abstract: Annotation watermarking is an application of watermarking where information about a cover or a part thereof are embedded in the cover itself to link both directly together. In earlier work we introduced Nested Object Annotation Watermarking as a special case, where the semantic, shape and hierarchical relations between the depicted nested objects are embedded in the area of each object only. As these regions can be anywhere and may be composed of any shape there is very limited a-priori knowledge for synchronization, which results in a higher complexity and ultimately in a higher error-proneness. In general an exhaustive search strategy for the proper blocks to reconstruct the shape suffers from the intrinsic combinatorial explosion of this process. Therefore in earlier work, at first we focused on rectangular embedding schemes with a block luminance algorithm, a steganographic WetPaperCode algorithm, a rectangular and finally a polygonal Dual-Domain DFT algorithm. In this paper we review and compare these algorithms in terms of their transparency, capacity, shape fidelity and robustness against the selected aspects JPEG compression and cropping. For the DFT algorithm we also show the influence of several parameters, present our idea of a method to reduce the combinatorial explosion by collating the paths in the search tree, and show that our new synchronization approach surpasses our former rectangular method in terms of correct retrievals, despite its higher complexity.

Der Einfluss gleichgewichteter Fusion in der Mikrofonforensik unter beispielhafter Nutzung von zwei Klassifikatoren

Abstract: Fur den beispielhaft gewahlten Anwendungsbereich der Mikrofonforensik wird in diesem Beitrag gezeigt, dass bisherig verwendete statistische Mustererkennungsverfahren zur Mikrofonerkennung von der Informationsfusion profitieren konnen. Dies wird im Beitrag mit Ergebnissen fur Fusionen auf Match-, Rankund Decision-Level sowie der Nutzung zweier Mehrklassenklassifikatoren (einem Entscheidungsbaumverfahren und logistischen Regressionsmodellen) belegt. Durch die Fusion konnten die erzielten Klassifikationsgenauigkeiten in der uberwachten Klassifikation auf den zwei hier genutzten Testmengen (eine mit vier und die anderer mit sieben Mikrofonen) zum Teil bis auf 100% erhoht werden. Damit wird der Wert der Mikrofonforensik als Methode der Quellverifikation z.B. fur die Langzeitarchivierung, weitergesteigert.

Evaluation of Fusion for Similarity Searching in Online Handwritten Documents

Abstract: With the spread of TabletPCs handwriting raises in its significance and importance in the digital domain. Also there exist other devices with pen-based inputs like PDAs, digitizer tablets and pads specially prepared with sensors. The advantage of handwritten input methods is their possibility of an ad hoc creation of technical sketches and drawings alongside with text and that keyboards may be in some cases and environments bothersome. Therefore the amount of handwritten documents is likely to increase. But a great problem is a proper full text search on such documents. This paper discusses the effects of multi-sample and multi-algorithm fusion approaches, known from biometrics to increase the performance. The tests are done by using three different devices (Logitech ioPen, Pegasus PC NotesMaker, ACE CAD DigiMemo Digital) and five different feature extraction methods (square grid, triangular grid, slope, curvature and slant of writing) and show that fusion can improve the retrieval performance in terms of precision and recall from 0.903 and 0.935 without fusion to 0.958 and 0.943 with fusion, respectively.