Multimedia Systems Conference 2012 ( MMSys '12). New York: ACM, 2012, pp. 23-28

3D Object Retrieval is an important field of research with many application possibilities. One of the main goals in this research is the development of discriminative methods for similarity search. The descriptor-based approach to date has seen a lot of research attention, with many different extraction algorithms proposed. In previous work, we have introduced a simple but effective scheme for 3D model retrieval based on a spatially fixed combination of 3D object fragment descriptors. In this work, we propose a novel flexible combination scheme based on finding the best matching fragment descriptors to use in the combination. By an exhaustive experimental evaluation on established benchmark data we show the capability of the new combination scheme to provide improved retrieval effectiveness. The method is proposed as a versatile and inexpensive method to enhance the effectiveness of a given global 3D descriptor approach.

Bruckstein, Alfred M. (Ed.) et al.: Scale Space and Variational Methods in Computer Vision. Proceedings : SSVM 2011. Berlin, Heidelberg, New York: Springer, 2012. (Lecture Notes in Computer Science (LNCS) 6667), pp. 62-73

We present an efficient implementation of volumetric nonlinear anisotropic image diffusion on modern programmable graphics processing units (GPUs). We avoid the computational bottleneck of a time consuming eigenvalue decomposition in R3. Instead, we use a projection of the Hessian matrix along the surface normal onto the tangent plane of the local isodensity surface and solve for the remaining two tangent space eigenvectors. We derive closed formulas to achieve this resulting in efficient GPU code. We show that our most complex volumetric nonlinear anisotropic diffusion gains a speed up of more than 600 compared to a CPU solution.

Thulasiraman, Parimala (Ed.) et al.: Proceedings 2011 IEEE International Conference on High Performance Computing and Communications : HPCC 2011. Los Alamitos, Calif.: IEEE Computer Society Conference Publishing Services (CPS), 2011, pp. 78-85

A personal computer can be considered as a one-node heterogeneous cluster that simultaneously processes several application tasks. It can be composed by, for example, asymmetric CPU and GPUs. This way, a high-performance heterogeneous platform is built on a desktop for data intensive engineering calculations. In our perspective, a workload distribution over the Processing Units (PUs) plays a key role in such systems. This issue presents challenges since the cost of a task at a PU is non-deterministic and can be affected by parameters not known a priori. This paper presents a context-aware runtime and tuning system based on a compromise between reducing the execution time of engineering applications - due to appropriate dynamic scheduling - and the cost of computing such scheduling applied on a platform composed of CPU and GPUs. Results obtained in experimental case studies are encouraging and a performance gain of 21.77% was achieved in comparison to the static assignment of all tasks to the GPU.

Smith, Michael J. (Ed.) et al.: Human Interface and the Management of Information: Part I : Interacting with Information. Berlin, Heidelberg, New York: Springer, 2011. (Lecture Notes in Computer Science (LNCS) 6771), pp. 528-537

Existing reference models for recommender systems are on an abstract level of detail or do not point out the processes and transitions of recommendation systems. However, this information is relevant for developers to design or improve recommendation systems. Even so, users need some background information of the calculation process to understand the process and accept or configure these systems proper. In this paper we present a comprehensive reference model for recommender systems which conjuncts the recommendation processes on an adequate level of detail. To achieve this, the processes of content-based and collaboration-based systems are merged and extended by the transitions and phases of hybrid systems. Furthermore, the algorithms which can be applied in the phases of the model are examined to identify the data flow between these phases. With our model those information of the recommendation calculation process can be identified, which encourages the traceability and thus the acceptance of recommendations.

Stephanidis, Constantine (Ed.): Universal Access in Human-Computer Interaction: Part I : Design for All and eInclusion. Berlin, Heidelberg, New York: Springer, 2011. (Lecture Notes in Computer Science (LNCS) 6765), pp. 20-29

Nowadays a wide range of input devices are available to users of technical systems. Especially modern alternative interaction devices, which are known from game consoles etc., provide a more natural way of interaction. But the support in computer programs is currently a big challenge, because a high effort is to invest for developing an application that supports such alternative input devices. For this fact we made a concept for an interaction system, which supports the use of alternative interaction devices. The interaction-system consists as central element a server, which provides a simple access interface for application to support such devices. It is also possible to address an abstract device by its properties and the interaction-system overtakes the converting from a concrete device. For realizing this idea, we also defined a taxonomy for classifying interaction devices by its interaction method and in dependence to the required interaction results, like recognized gestures. Later, by using this system, it is generally possible to develop a user-centered system by integrating this interaction-system, because an adequate integration of alternative interaction devices provides a more natural and easy to understand form of interaction.

Stephanidis, Constantine (Ed.): Universal Access in Human-Computer Interaction: Part I : Design for All and eInclusion. Berlin, Heidelberg, New York: Springer, 2011. (Lecture Notes in Computer Science (LNCS) 6765), pp. 317-326

Searching information on web turned to a matter of course in the last years. The visualization and filtering of the results of such search queries plays a key-role in different disciplines and is still today under research. In this paper a new approach for classifying the search intention of users' is presented. The approach uses existing and easy parameters for a differentiation between explorative and targeted search. The results of the classification are used for a differentiated presentation based on graphical visualization techniques.

Kurosu, Masaaki (Ed.): Human Centered Design : Second International Conference: HCD 2011. Berlin, Heidelberg, New York: Springer, 2011. (Lecture Notes in Computer Science (LNCS) 6776), pp. 491-499

The Semantic Web is a powerful technology for organizing the data in our information based society. The collection and organization of information is an important step for showing important information to interested people. But the usage of such semantic-based data sources depends on effective and efficient information visualizations. Currently different kinds of visualizations in general and visualization metaphors do exist. Many of them are also applied for semantic data source, but often they are designed for semantic web experts and neglecting the normal user and his perception of an easy useable visualization. This kind of user needs less information, but rather a reduced qualitative view on the data. These two aspects of large amount of existing data and one for normal users easy to understand visualization is often not reconcilable. In this paper we create a concept for a visualization to show a bigger set of information to such normal users without overstraining them, because of layer-based data visualization, next to an integration of a Focus and Context metaphor.

Moderne Interaktionsgeräte orientieren sich bei der Interaktion zunehmend an der natürlich menschlichen Interaktion, wie beispielsweise über Gesten. Dieser Trend begann 2006 insbesondere durch den Verkaufserfolg der Spielekonsole Wii von Nintendo und dem dazugehörigen gestenbasierten Controller, der WiiMote. Hierbei war es erstmals möglich Spiele komplett über Gesten zu bedienen. Der Vorteil dieser modernen Interaktion ist die einfachere und leicht verständlichere Kommunikation zwischen und Mensch und Maschine. Im Gegensatz zu solchen proprietären Systemen, die mittels speziell aufeinander abgestimmten Hardware und Software arbeiten, sind im PC Bereich gestenbasierte Steuerungen selten verwendet. Der Grund liegt in einer fehlenden zentralen Unterstützung, so dass Anwendungsentwickler solche Geräte für jede Anwendung individuell berücksichtigen müssen. Dies stellt aber einen großen zusätzlichen Aufwand dar, der selten im Rahmen der Softwareentwicklung berücksichtigt wird. Aus diesem Grund gibt es auch nur wenige Umsetzungen von Anwendungen, die alternative Eingabegeräte unterstützen. Zumeist handelt es sich bei solchen Anwendungen um Spezialsoftware, für deren Zweck die Eingabegeräte und Einbindung der Geräte optimiert sind. Die vorliegende Arbeit sieht daher als Ziel die Entwicklung eines Konzepts für eine zentrale Unterstützung moderner alternativer Eingabegeräte. Hierfür muss zusätzlich eine Klassifikation entwickelt werden, die eine Organisation der verschiedenen Eingabegeräte unter Berücksichtigung der Interaktionsmethoden und benötigtem Interaktionsergebnis (z.B. Koordinaten, erkannte Geste) erlaubt und so eine spätere Einbindung in eine Anwendung vereinfacht und fördert. Mittels dieses Konzepts ist es möglich ein System zu entwickeln, das alle Geräte und deren Interaktionsformen organisiert. Ein Entwickler ist damit in der Lage, unter Verwendung der bereitgestellten Schnittstellen, diese Eingabegeräte in seine Anwendung einzubinden und sie als Kommunikationsmittel für den Nutzer bereitstellen. Zur Entwicklung eines solchen Systems werden zunächst theoretische Ansätze aus dem Bereich der Mensch-Computer-Interaktion betrachtet, die als Grundlage für die methodische Anwendung dienen soll. Dies umfasst auch eine Betrachtung aktuell verfügbarer alternativer Eingabegeräte. Ausgehend von den gewonnenen Erkenntnissen werden insbesondere aktuelle Klassifikationskonzepte, sowie Umsetzungen von Interaktionssystemen betrachtet. Die gesammelten Erkenntnisse werden herangezogen, um ein Konzept für ein Interaktionssystem zu entwickelt und deren zugrundeliegenden Organisation der zu unterstützenden Eingabegeräte. Auf Basis dieses Konzeptes folgt abschließend eine Beschreibung der prototypisch technischen Umsetzung, unter der beispielhaften Verwendung der WiiMote.

Richard, Paul et al.: Computer Vision, Imaging and Computer Graphics. Theory and Applications : International Joint Conference, VISIGRAPP 2010. Berlin, Heidelberg, New York: Springer, 2011. (Communications in Computer and Information Science 229), pp. 96-109

Segmenting foreground from background automatically is an active field of research. The graph cut approach is one of the promising methods to solve this problem. This approach requires that the weights of the graph are chosen optimally in order to obtain a good segmentation. We address this challenge focusing on the automatic segmentation of wood log images. We present a novel method based on density estimation to obtain information about both foreground and background. With this information the weights in the graph cut method can be set automatically. In order to validate our results, we use four different methods to set these weights. We show that of these approaches, our new method obtains the best results.

Bender, Jan (Ed.) et al.: VRIPHYS 11: 8th Workshop in Virtual Reality Interactions and Physical Simulations. Goslar: Eurographics Association, 2011, pp. 73-82

A natural interaction with virtual environments is one of the key issues for the usability of Virtual Reality applications. Device-free, intuitive interactions with the virtual world can be achieved by capturing the movements of the user with markerless motion capture. In this work we present a markerless motion capture approach which can be used to estimate the human body pose in real-time with a single depth camera. The presented approach requires neither a 3D shape model of the tracked person nor a training phase in which body shapes are learned a priori. Instead, it analyzes the curvature of the human body to estimate the symmetry axes of the body joints. These symmetry axes are then used to calculate the pose of the tracked human in real-time. The presented approach was evaluated qualitatively with a time-of-flight and a Kinect depth camera. Furthermore, quantitative simulation results show that the proposed approach is promising for depth cameras which can reliably capture the surface curvature (and thus the normals) of a person and which have a resolution of at least 320x240 pixel.

In dieser Arbeit wird ein neuer Ansatz im Bereich des markerlosen Motion Capturing vorgestellt, der die Bewegung der Arme, Beine und des Oberkörpers erfasst. Die Erfassung der Bewegung erfolgt dabei in Echtzeit aus den 3D-Messdaten einer Tiefenkamera. Die Tiefendaten sind unvollständig, da die Kamera die Szene nur aus einer Blickrichtung erfassen kann. Der Algorithmus nutzt die zylinderähnliche Form der Arme, Beine und des Oberkörpers aus und berechnet aus den Tiefenbildern die mediale Achse der Körperteile. Für eine effiziente Bestimmung der medialen Achsen werden die räumlichen Nachbarschaftsinformationen der Tiefenbilder ausgenutzt. Die Bewegung wird auf ein virtuelles Skelett abgebildet, indem die Knochen entsprechend der medialen Achsen ausgerichtet werden. Die Bestimmung der Pose erfolgt von Bild zu Bild, so dass eine berechnete Pose als zusätzliche Eingabe für das nächste Bild dient.

IEEE Computer Society: XIII Symposium on Virtual Reality : SVR 2011. Los Alamitos, Calif.: IEEE Computer Society Conference Publishing Services (CPS), 2011, pp. 1-8

Over the years, faster hardware - with higher clock rates - has been the usual way to improve computing times in computer graphics. Aside from highly costly parallel solutions only affordable by big industries - like the movie industry -, there was no alternative available to desktop users. Nevertheless, this scenario is dramatically changing with the introduction of more and more parallelism in current desktop PCs. Multi-core CPUs are a common basis in current PCs and the power of modern GPUs - which have been multi-core for a long time now - is getting unveiled to developers. nVidia's CUDA is a powerful weapon to explore GPUs parallelism. Yet, its specific target - nVidia graphic cards only - does not provide any solution to other parallel hardware present. OpenCL is a new royalty-free cross-platform intended to be portable across different hardware manufacturers or even different platforms. In this paper we focus on a comparison of advantages and disadvantages of xPU platforms with OpenCL and CUDA in terms of time efficiency. As an example application we use ray tracing algorithms. Three kinds of ray tracers have to be developed in order to conduct a fair comparison: one is CPU based, while the other two are GPU based - using CUDA and OpenCL, respectively. At the end, a comparison is done between them and results are presented and analyzed showing that the CUDA implementation has the best frame rate, but is very closely followed by the OpenCL implementation. Visually, results are identical, showing the high potential of OpenCL as an alternative for CUDA with identical performance.

Shumaker, Randall (Ed.): Virtual and Mixed Reality - Systems and Applications: Part II : International Conference, Virtual and Mixed Reality 2011. Berlin, Heidelberg, New York: Springer, 2011. (Lecture Notes in Computer Science (LNCS) 6774), pp. 343-351

In this paper we present a GPU-accelerated implementation of the well-known freeform deformation algorithm to allow for deformable objects within fully interactive virtual environments. We furthermore outline how our real-time deformation approach can be integrated into the X3D standard for more accessibility of the proposed methods. The presented technique can be used to deform complex detailed geometries without pre-processing the mesh by simply generating a lattice around the model. The local deformation is then computed for this lattice instead of the complex geometry, which efficiently can be carried out on the GPU using CUDA.

John, Nigel (Co-Chair) et al.: Eurographics 2011. State of the Art Reports (STARs). Eurographics Association, 2011, pp. 75-100

For many application areas, where a task is most naturally represented by talking or where standard input devices are difficult to use or not available at all, virtual characters can be well suited as an intuitive man-machine-interface due to their inherent ability to simulate verbal as well as nonverbal communicative behavior. This type of interface is made possible with the help of multimodal dialog systems, which extend common speech dialog systems with additional modalities just like in human-human interaction. Multimodal dialog systems consist at least of an auditive and graphical component, and communication is based on speech and nonverbal communication alike. However, employing virtual characters as personal and believable dialog partners in multimodal dialogs entails several challenges, because this requires not only a reliable and consistent motion and dialog behavior but also regarding nonverbal communication and affective components. Besides modeling the "mind" and creating intelligent communication behavior on the encoding side, which is an active field of research in artificial intelligence, the visual representation of a character including its perceivable behavior, from a decoding perspective, such as facial expressions and gestures, belongs to the domain of computer graphics and likewise implicates many open issues concerning natural communication. Therefore, in this report we give a comprehensive overview how to go from communication models to actual animation and rendering.

Richard, Paul (Ed.) et al.: GRAPP 2011 : Proceedings of the International Conference on Computer Graphics Theory and Applications. Portugal: SciTePress, 2011, pp. 286-295

Multi-projector systems are widely used in many application areas. Such systems are for instance employed to increase the brightness or the resolution of projected images. Intrinsic to multi-projector systems are problems like geometric misalignment, especially when projecting onto complex arbitrarily formed projection surfaces, and photometric deviations. Therefore, several difficult calibration tasks (geometry, brightness, color) have to be performed. A high-quality and easy-to-use calibration process is the key to good usability for untrained or unexperienced users. Due to the fact that manual calibration is time-consuming and imprecise, automatic approaches were developed in recent years. This paper analyzes the most popular state-of-the-art algorithms and setups with respect to their advantages and disadvantages. We summarize the general working principles of calibration algorithms and provide an outlook into the fields in which the described algorithms are most useful.

IAPR Newsletter, Vol.33 (2011), 1, pp. 22-25

DAGM is the German section of the IAPR. Every year a conference is held in Germany (or one of the surrounding countries). This year, DAGM took place in Darmstadt. The main conference was preceded by a day at Fraunhofer IGD with a workshop on "Pattern Recognition for IT Security".

Computer Graphics Forum, Vol.30 (2011), 6, pp. 1719-1749

The analysis of large graphs plays a prominent role in various fields of research and is relevant in many important application areas. Effective visual analysis of graphs requires appropriate visual presentations in combination with respective user interaction facilities and algorithmic graph analysis methods. How to design appropriate graph analysis systems depends on many factors, including the type of graph describing the data, the analytical task at hand and the applicability of graph analysis methods. The most recent surveys of graph visualization and navigation techniques cover techniques that had been introduced until 2000 or concentrate only on graph layouts published until 2002. Recently, new techniques have been developed covering a broader range of graph types, such as timevarying graphs. Also, in accordance with ever growing amounts of graph-structured data becoming available, the inclusion of algorithmic graph analysis and interaction techniques becomes increasingly important. In this State-of-the-Art Report, we survey available techniques for the visual analysis of large graphs. Our review first considers graph visualization techniques according to the type of graphs supported. The visualization techniques form the basis for the presentation of interaction approaches suitable for visual graph exploration. As an important component of visual graph analysis, we discuss various graph algorithmic aspects useful for the different stages of the visual graph analysis process. We also present main open research challenges in this field.

Smith, Michael J. (Ed.) et al.: Human Interface and the Management of Information: Part I : Interacting with Information. Berlin, Heidelberg, New York: Springer, 2011. (Lecture Notes in Computer Science (LNCS) 6771), pp. 567-576

Ambient Assisted Living (AAL) comprises methods, systems, and services applied to improve the quality of daily life for humans, especially elderly people. Recent research emphasizes the implementation of comprehensive AAL platforms which control all technological components included in the entire environment such as one's apartment. The behavior of the system is often determined by a specific set of rules. Thus, personalization according to the person's needs and preferences includes a configuration of the given rule system. Assuming that configuration is not only conducted by technical staff but also by the person him or herself, this process can be regarded as complex, requiring technical knowledge. In this work, we present an interactive and architectural approach to support at the personalization of an AAL system by different types of users.

Smith, Michael J. (Ed.) et al.: Human Interface and the Management of Information: Part I : Interacting with Information. Berlin, Heidelberg, New York: Springer, 2011. (Lecture Notes in Computer Science (LNCS) 6771), pp. 239-248

Semantically annotated data gain more and more importance in future information acquiring processes. Especially the Linked Open Data (LOD) format has already experienced a great growth. However, the user-interfaces of web-applications mostly do not reflect the added value of semantics data. The following paper describes a new approach of user-centered data-adaptive semantics visualization, which makes use of the advantages of semantics data combined with an adaptive composition of information visualization techniques. It starts with a related work section, where existing LOD systems and information visualization techniques are described. After that, the new approach will bridge the gap between semantically annotated data (LOD) and information visualization and introduces a visualization system that adapts the composition of visualizations based on the underlying data structure. A case study of an example case will conclude this paper.

Kurosu, Masaaki (Ed.): Human Centered Design : Second International Conference: HCD 2011. Berlin, Heidelberg, New York: Springer, 2011. (Lecture Notes in Computer Science (LNCS) 6776), pp. 576-585

Presenting information in a user-oriented way has a significant impact on the success and comprehensibility of data visualizations. In order to correctly and comprehensibly visualize data in a user-oriented way data specific aspects have to be considered. Furthermore, user-oriented perception characteristics are decisive for the fast and proper interpretation of the visualized data. In this paper we present a taxonomy for graph visualization techniques. On the one hand it provides the user-oriented identification of applicable visual features for given data to be visualized. On the other hand the set of visualization techniques is enclosed which supports these identified visual features. Thus, the taxonomy supports the development of user-oriented visualizations by examination of data to obtain a beneficial association of data to visual features.

Stephanidis, Constantine (Ed.): Universal Access in Human-Computer Interaction: Part II : Users Diversity. Berlin, Heidelberg, New York: Springer, 2011. (Lecture Notes in Computer Science (LNCS) 6766), pp. 88-97

The automatic adaptation of information visualization systems to the requirements of users plays a key-role in today's research. Different approaches from both disciplines try to face this phenomenon. The modeling of user is an essential part of a user-centered adaptation of visualization. In this paper we introduce a new approach for modeling users especially for semantic visualization systems. The approach consists of a three dimensional model, where semantic data, user and visualization are set in relation in different abstraction layer.

Kurosu, Masaaki (Ed.): Human Centered Design : Second International Conference: HCD 2011. Berlin, Heidelberg, New York: Springer, 2011. (Lecture Notes in Computer Science (LNCS) 6776), pp. 566-575

Today the information visualization takes in an important position, because it is required in nearly every context where large databases have to be visualized. For this challenge new approaches are needed to allow the user an adequate access to these data. Static visualizations are only able to show the data without any support to the users, which is the reason for the accomplished researches to adaptive user-interfaces, in particular for adaptive visualizations. By these approaches the visualizations were adapted to the users' behavior, so that graphical primitives were change to support a user e.g. by highlighting user-specific entities, which seems relevant for a user. This approach is commonly used, but it is limited on changes for just a single visualization. Modern heterogeneous data providing different kinds of aspects, which modern visualizations try to regard, but therefore a user often needs more than a single visualization for making an information retrieval. In this paper we describe a concept for adapting the user-interface by selecting visualizations in dependence to automatically generated data characteristics. So visualizations will be chosen, which are fitting well to the generated characteristics. Finally the user gets an aquatically arranged set of visualizations as initial point of his interaction through the data.

Jacko, Julie A. (Ed.): Human-Computer Interaction: Part I : Design and Development Approaches. Berlin, Heidelberg, New York: Springer, 2011. (Lecture Notes in Computer Science (LNCS) 6761), pp. 480-489

One key issue of both Information Visualization as well as Adaptive User Interfaces is information overload. While both disciplines have already devised well performing algorithms, methods and applications, a real merging has not taken place yet. Only a few attempts bring the surplus values of both disciplines together, whereas a fine-grained investigation of visualization parameterization is not investigated. Today's systems focus either on the adaptation of visualization types or the parameterization of visualizations. This paper presents a reference Model for Adaptive Visualization Systems (MAVS) that allows the adaptation of both the visualization type and the visualization parameterization. Based on this model, a framework for the adaptive visualization of semantics data will be derived. A use case describing the interaction with an "adaptive visualization cockpit" covering different visualization metaphors concludes the paper.

Die 3D-Koordinaten von Punkten auf einer Objektoberfläche können entweder mit einer Tiefenkamera gemessen oder durch Structure-from-Motion Verfahren aus mehreren 2D Bildern rekonstruiert werden. In dieser Arbeit wird vergleichend evaluiert, welche Genauigkeit mit diesen beiden Verfahren erreicht werden kann. Für eine aussagekräftige Evaluierung muss die Kamerapose zur Berechnung der (echten) 3D-Referenzdaten bekannt sein. Während in vorhergehenden Ansätzen die Kamerapose bildbasiert (z.B. anhand eines Marker Trackers) geschätzt wurde, wird die Kamerapose in dieser Arbeit anhand eines hochpräzisen Faro-Messarms bestimmt. Hierdurch wird eine höhere Genauigkeit erreicht als bei Bestimmung der Kamerapose anhand eines Marker Trackers, was die Aussagekraft der Evalierungsergebnisse erhöht.

In a face or facial expression recognition system, one of the key requirements is the proper generation of a face model. An approach for reconstructing the 3D face model from a video sequence is presented within this thesis. A generic deformable 3D face model is built from the 3D scans of one hundred individuals. This generic 3D face model allows modelling the variation in the face-shape of different people. It can be fitted to the reconstructed 3D face-shape of an individual from a video sequence provided that the deviation from the real 3D face is less than certain thresholds. The methodology for reconstructing the 3D face-shape from a video sequence is provided. A structure-from-motion technique is used to estimate the poses of images in a video sequence. The estimated poses are optimized further by applying a bundle adjustment technique. Finally the dense correspondences between the images are estimated by employing Huber-L1 optical flow algorithm. With the optimized pose parameters and dense correspondences between the images, dense 3D face-shapes are reconstructed from a video sequence. The application is developed to reconstruct the 3D face-shape in nearly uncontrolled environment where challenges exists due to varying lightning conditions, the non steady camera movement while capturing a video sequence and other factors using most promising state-of-the-art algorithm. Given the nearly uncontrolled conditions we were able to estimate a point cloud representing the face shape, but the results are not as accurate as expected. The related state-of-the-art algorithms used here have not been considered for such challenging conditions. The factors affecting the depth estimation in face region are analyzed and possible improvements to enhance the 3D face-shape reconstruction are presented within in this thesis.

IEEE Computer Society: Proceedings of the 2011 IEEE 1st International Network Science Workshop : NSW 2011. Printing House, 2011, pp. 67-74

Connectivity and security of tactical mobile ad hoc networks (MANETs) can be enhanced significantly by explicit consideration of radio propagation as this not only allows determination of route feasibility and minimization of radio frequency power, but also avoids detectability of emanations. Tactical MANETs (and increasingly general MANETs) typically have geolocation and terrain information available, however as they are likely to be deployed in urban, broken, or indoor environments, simple Free Space or Two-Ray Ground models are not adequate. Although highly accurate radio propagation models exist, they require considerable computational resources and are hence unsuitable for incorporation into real-time protocols, particularly on resource-constrained platforms such as MANET nodes. We therefore propose a simplified, scalable ray-optical radio frequency propagation model that incorporates a Two-Ray Ground model and takes reflections and deflections on terrain features into account. Although our proposed model does not incorporate a comprehensive model of all physical effects, we argue that the approximation provided by our model is sufficient and suitable for the purposes of enhancing network performance and accuracy in the frequency range currently used by wireless networks. The model was incorporated in the NS-2 simulator and validated both using simulation and experimentally.

Diese Arbeit gibt einen Einblick in die Grundlagen der 3D-Anwendungen im Zusammenspiel mit verschiedenen Customer Devices. In diesem Fall ist die Anwendung ein 3D-Planetarium unseres Sonnensystems für ein Museum. Ziel ist es, ein Konzept zur intuitiven und immersiven Navigation durch das virtuelle Planetarium mittels preiswerte Customer Devices zu erstellen. Die Besucher sollen sich frei und ohne großen Aufwand im Sonnensystem bewegen können. Dabei soll sich der Besucher auf die Simulation konzentrieren können und nicht durch komplexe Steuerung schnell das Interesse an der Anwendung verlieren. Dafür werden vergleichbare Ansätze und der bisherige Forschungsstand untersucht und ein neuer Ansatz beschrieben. Als kostengünstige Customer Devices werden in dieser Arbeit die Controller der Nintendo Wii (Wiimote) und aktuelle Smartphones betrachtet. Eine genaue Analyse dieser Geräte ist Bestandteil dieser Arbeit. Ausgehend von den gewählten Geräten gibt es verschiedene Interaktionsmöglichkeiten und daraus resultierende Interaktionskonzepte. Für jedes Gerät soll ein Konzept entwickelt werden, das die gestellten Ansprüche erfüllt.

Jacko, Julie A. (Ed.): Human-Computer Interaction: Part IV : Users and Applications. Berlin, Heidelberg, New York: Springer, 2011. (Lecture Notes in Computer Science (LNCS) 6764), pp. 520-529

Time appears in many different semantic information systems like historical databases, multimedia systems or social communities as a common attribute. Beside the temporal information, the resources in these domains are categorized in a domain-specific schema and interconnected by semantic relations. Nevertheless, the high potential of these systems is not yet exhausted completely. Even today most of these knowledge systems present timedependent semantic knowledge in textual form, what makes it difficult for the average user to understand temporal structures and dependencies. For bridging this gap between human and computer and for simplifying the exploration of time-dependent semantic knowledge, we developed a new interactive timeline visualization called SemaTime. The new designed temporal navigation concept offers an intuitive way for exploring and filtering time-depended resources. Additionally SemaTime offers navigation and visual filtering methods on the conceptual layer of the domain and is able to depict semantic relations. In this paper we describe the conceptual design of SemaTime and illustrate its application potentials in semantic search environments.

Computers & Graphics, Vol.35 (2011), 4, pp. 778-788

This paper focuses on the preparative process of natural feature map retrieval for a mobile camera-based tracking system. We cover the most important aspects of a general purpose tracking system including the acquisition of the scene's geometry, tracking initialization and fast and accurate frame-by-frame tracking. To this end, several state-of-the-art techniques - each targeted at one particular subproblem - are fused together, whereby their interplay and complementary benefits form the core of the system and the thread of our discussion. The choice of the individual sub-algorithms in our system reflects the scarcity of computational resources on mobile devices. In order to allow a more accurate, more robust and faster tracking during run-time, we therefore transfer the computational load into the preparative customization step wherever possible. From the viewpoint of the user, the preparative stage is kept very simple. It only involves recording the scene from various viewpoints and defining a transformation into a target coordinate frame via manual definition of only a few 3D to 3D point correspondences. Technically, the image sequence is used to (1) capture the scene's geometry by a SLAM-Method and subsequent refinement via constrained Bundle Adjustment, (2) to train a Randomized-Trees classifier for wide-baseline tracking initialization, and (3) to analyze the view-point dependent visibility of each feature. During run-time, robustness and performance of the frame-to-frame tracking are further increased by fusing inertial measurements within a combined pose estimation.

IEEE Computer Society: 3DIMPVT 2011 : International Conference on 3D Imaging, Modeling, Processing, Visualization and Transmission. Los Alamitos, Calif.: IEEE Computer Society Conference Publishing Services (CPS), 2011, pp. 140-147

This paper focuses on the preparative process of retrieving accurate feature maps for a camera-based tracking system. With this system it is possible to create ready-touse Augmented Reality applications with a very easy setup work-flow, which in practice only involves three steps: filming the object or environment from various viewpoints, defining a transformation between the reconstructed map and the target coordinate frame based on a small number of 3D-3D correspondences and, finally, initiating a feature learning and Bundle Adjustment step. Technically, the solution comprises several sub-algorithms. Given the image sequence provided by the user, a feature map is initially reconstructed and incrementally extended using a Simultaneous-Localization-and-Mapping (SLAM) approach. For the automatic initialization of the SLAM module, a method for detecting the amount of translation is proposed. Since the initially reconstructed map is defined in an arbitrary coordinate system, we present a method for optimally aligning the feature map to the target coordinated frame of the augmentation models based on 3D-3D correspondences defined by the user. As an initial estimate we solve for a rigid transformation with scaling, known as Absolute Orientation. For refinement of the alignment we present a modification of the well-known Bundle Adjustment, where we include these 3D-3D-correspondences as constraints. Compared to ordinary Bundle Adjustment we show that this leads to significantly more accurate reconstructions, since map deformations due to systematic errors such as small camera calibration errors or outliers are well compensated. This again results in a better alignment of the augmentations during run-time of the application, even in large-scale environments.

Loncaric, Sven (Ed.) et al.: ISPA 2011 : 7th International Symposium on Image and Signal Processing and Analysis. Zagreb: University of Zagreb, 2011, pp. 437-442

The number of available 3D models in various areas increases steadily. Efficient methods to search for 3D models by content, rather than textual annotations, are crucial. For this purpose, we propose content based 3D model retrieval using a compressive sensing technique which is very efficient in classification by using only few input information. Our approach to search and automatically return a set of 3D mesh models from a large database consists of three major steps: (1) suggestive contours extraction from different viewpoints to extract features of the query 3D model; (2) descriptor computation by analyzing the Histogram of Oriented Gradients of the suggestive contours in the space of diffusion tensor fields; and (3) compressive sensing based machine learning to retrieve the models and the most probable view-point. Experimental results show that our proposed 3D model retrieval system is very effective to retrieve the 3D models, even though there are variations of shape and pose of the models.

Bebis, George (Ed.) et al.: Advances in Visual Computing. 7th International Symposium, ISVC 2011 : Proceedings, Part II. Berlin, Heidelberg, New York: Springer, 2011. (Lecture Notes in Computer Science (LNCS) 6939), pp. 367-376

The number of available 3D models in various areas increases steadily. Efficient methods to search for 3D models by content, rather than textual annotations, are crucial. For this purpose, we propose a content based 3D model retrieval system using the Histogram of Orientation (HoO) from suggestive contours and their diffusion tensor fields. Our approach to search and automatically return a set of 3D mesh models from a large database consists of three major steps: (1) suggestive contours extraction from different viewpoints to extract features of the query 3D model; (2) HoO descriptor computation by analyzing the diffusion tensor fields of the suggestive contours; (3) similarity measurement to retrieve the models and the most probable view-point. Our proposed 3D model retrieval system is very efficient to retrieve the 3D models even though there are variations of shape and pose of the models. Experimental results are presented and indicate the effectiveness of our approach, competing with the current - more complicated - state of the art method and even improving results for several classes.

Electronics Letters, Vol.47 (2011), 21, pp. 1181-1183

A sketch-based 3D model retrieval methodology by using compressive sensing (CS) is presented. The approach to search and automatically return a set of 3D mesh models from a large database consists of histogram of oriented gradient feature extraction from projected images of 3D models, and CS based classification technique. Experimental results show that the proposed 3D model retrieval is very efficient to search for the 3D models from user-drawn sketches including the variation of shape, pose, and view-points.

IEEE Computer Society: International Joint Conference on Biometrics 2011 : IJCB 2011. Los Alamitos, Calif.: IEEE Computer Society Conference Publishing Services (CPS), 2011, 8 p.

Fuzzy commitment is an efficient template protection algorithm that can improve security and safeguard privacy of biometrics. Existing theoretical security analysis has proved that although privacy leakage is unavoidable, perfect security from information-theoretical points of view is possible when bits extracted from biometric features are uniformly and independently distributed. Unfortunately, this strict condition is difficult to fulfill in practice. In many applications, dependency of binary features is ignored and security is thus suspected to be highly overestimated. This paper gives a comprehensive analysis on security and privacy of fuzzy commitment regarding empirical evaluation. The criteria representing requirements in practical applications are investigated and measured quantitatively in an existing protection system for 3D face recognition. The evaluation results show that a very significant reduction of security and enlargement of privacy leakage occur due to the dependency of biometric features. This work shows that in practice, one has to explicitly measure the security and privacy instead of trusting results under non-realistic assumptions.

IEEE Computer Society: Proceedings 2010 12th IEEE International Conference on High Performance Computing and Communications : HPCC 2010. Los Alamitos, Calif.: IEEE Computer Society Conference Publishing Services (CPS), 2010, pp. 305-313

GPUs (Graphics Processing Units) have become one of the main co-processors that contributed to desktops towards high performance computing. Together with multi-core CPUs, a powerful heterogeneous execution platform is built for massive calculations. To improve application performance and explore this heterogeneity, a distribution of workload in a balanced way over the PUs (Processing Units) plays an important role for the system. However, this problem faces challenges since the cost of a task at a PU is non-deterministic and can be influenced by several parameters not known a priori, like the problem size domain. We present a comparison of iterative SLE (Systems of Linear Equations) solvers, used in many scientific and engineering applications, over a heterogeneous CPU-GPUs platform and characterize scenarios where the solvers obtain better performances. A new technique to improve memory access on matrix vector multiplication used by SLEs on GPUs is described and compared to standard implementations for CPU and GPUs. Such timing profiling is analyzed and break-even points based on the problem sizes are identified for this implementation, pointing whether our technique is faster to use GPU instead of CPU. Preliminary results show the importance of this study applied to a real-time CFD (Computational Fluid Dynamics) application with geometry modification.

Bentes, Cristiana (Ed.) et al.: SBAC-PADW 2010 : 1st Workshop on Applications for Multi and Many Core Architectures (WAMMCA). Los Alamitos, Calif.: IEEE Computer Society Conference Publishing Services (CPS), 2010, pp. 37-42

GPUs (Graphics Processing Units) have become one of the main co-processors that contributed to desktops towards high performance computing. Together with multicore CPUs and other co-processors, a powerful heterogeneous execution platform is built on a desktop for data intensive calculations. In our perspective, we see the modern desktop as a heterogeneous cluster that can deal with several applications' tasks at the same time. To improve application performance and explore such heterogeneity, a distribution of workload over the asymmetric PUs (Processing Units) plays an important role for the system. However, this problem faces challenges since the cost of a task at a PU is non-deterministic and can be influenced by several parameters not known a priori, like the problem size domain. We present a context-aware architecture that maximizes application performance on such platforms. This approach combines a model for a first scheduling based on an offline performance benchmark with a runtime model that keeps track of tasks' real performance. We carried a demonstration using a CPU-GPU platform for computing iterative SLEs (Systems of Linear Equations) solvers using the number of unknowns as the main parameter for assignment decision. We achieved a gain of 38.3% in comparison to the static assignment of all tasks to the GPU (which is done by current programming models, such as OpenCL and CUDA for Nvidia).

Fifth International Symposium on 3D Data Processing, Visualization and Transmission. Proceedings [online]. [cited 17 January 2011] Available from: http://campwww.informatik.tu-muenchen.de/3DPVT2010/data/media/e-proceeding/index.html 2010, 8 p.

We present a method to improve surface reconstruction using color encoded structured light. A video camera is used to obtain a color encoded pattern projected onto the object using a projector. A multi-stage method is presented to reconstruct the surface from the captured pattern. It consists of i) a robust edge detection step, ii) a color decoding using feedback from previous stripes and iii) a propagation step to detect errors and propagate detected stripes. Using feedback and propagation we can reconstruct areas where regular color detection fails by bridging small gaps which happen due to illumination variations or noise. The reconstruction results are illustrated for human skin.

Signal Processing (IIH-MSP-2010), it is our pleasure to present to you the conference proceedings. Multimedia technologies with an increasing level of intelligence are emerging to enable the process of creating a global information infrastructure that will interconnect heterogeneous computer networks and various forms of information technologies around the world. IIHMSP-2010, which is the sixth edition of the IIHMSP series of conferences, is intended as an international forum for researchers and professionals in all area of information hiding and multimedia signal processing. The major areas covered at the conference and presented in the proceedings include: · Information Hiding and Security · Multimedia Signal Processing and Networking · Bio-Inspired Multimedia Technologies and Systems

The technical program covered all aspects of pattern recognition and, to name only a few areas, ranged from 3D reconstruction, to object recognition and medical applications.

Institute for Systems and Technologies of Information, Control and Communication (INSTICC): VISIGRAPP 2010. Proceedings : International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications. INSTICC Press, 2010, pp. 60-67

The automatic extraction of foreground objects from the background is a well known problem. Much research has been done to solve the foreground/background segmentation with graph cuts. The major challenge is to determine the weights of the graph in order to obtain a good segmentation. In this paper we address this problem with a focus on the automatic segmentation of wood logs. We introduce a new solution to get information about foreground and background. This information is used to set the weights of the graph cut method. We compare four different methods to set these weights and show that the best results are obtained with our novel method, which is based on density estimation.

Computer Graphics Society (CGS): Computer Graphics International 2010. Short Papers : CGI [online]. [cited 01 February 2011] Available from: http://cgi2010.miralab.unige.ch/CGI_ShortPapersD4.html 2010, 4 p.

Despite the superior image quality generated by ray tracing, programmers of time-critical applications have historically avoided it because of its computational costs. Nowadays, the hardware of modern desktops allows the execution of realtime ray tracers but requires a specialized implementation based on specific characteristics of each application, such as scene complexity, kinds of motion, ray distribution, model structure and hardware. The evaluation and development of these requirements are complex and time-consuming, especially for developers with no familiarity in rendering algorithms and graphics hardware programming. The aim of our work is to provide a general and practical method to efficiently execute interactive ray tracing in most systems. We considered the most common aspects of current computer graphics applications, like the use of a scene graph and support to static and dynamic objects. In addition, we also took into account the common desktop hardware. This led us to the development of a special acceleration structure and its implementation on the GPU. In this paper, we present the development of our work showing the combination of different techniques and our results.

Brazilian Computer Society (SBC): XII Symposium on Virtual and Augmented Reality : SVR 2010. Brazil: Everton Cavalcante, 2010, 4 p.

For many years the Graphics Processing Unit (GPU) of common desktops was just used to accelerate certain parts of the graphics pipeline. After developers had access to the native instruction set and memory of the massive parallel computational elements of GPUs a lot has changed. GPUs became powerful and programmable. Nowadays two SDKs are most used for GPU programming: CUDA and OpenCL. CUDA is the most adopted general purpose parallel computing architecture for GPUs but is restricted to Nvidia graphic cards only. In contrast, OpenCL is a new royalityfree framework for parallel programming intended to be portable across different hardware manufacturers or even different platforms. In this paper, we evaluate both solutions considering a typical parallel algorithm: Ray Tracing. We show our performance results and experiences on developing both implementations that could be easily adapted to solve other problems.

Erleben, Kenny (Ed.) et al.: VRIPHYS 10: 7th Workshop in Virtual Reality Interactions and Physical Simulations. Goslar: Eurographics Association, 2010, pp. 21-29

In the field of garment simulation the resolution of the simulation mesh has a direct impact on visual quality. Unfortunately, an increase in mesh resolution introduces a much higher computational cost and potentially causes instability inside the simulation. In addition, it increases the amount of data sent to the renderer for visualisation. Therefore, a GPU-based refinement of the simulated mesh has several advantages, since all additional data is generated immediately before rendering. This allows an increase in visual quality without adding to computational costs for the simulation process or bandwidth necessary for rendering. In this paper we present a view-dependent, adaptive tessellation method designed for the geometry processing stage of modern GPUs. It uses uniform meshes internally, removing the necessity to store external patches. Since we deal with a local refinement scheme, sudden changes in the mesh structure size on adjacent patches may occur incidentally. To reduce this effect as far as possible, we control the triangle density distribution of the refinement process inside a refined triangle patch.

Institute for Systems and Technologies of Information, Control and Communication (INSTICC): VISIGRAPP 2010. Proceedings : International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications. INSTICC Press, 2010, pp. 246-253

Scene graphs are widely used as a description of spatial relations between objects in a scene. Current scene graphs use linear transformations for this purpose. This limits the relation of two objects in the hierarchy to simple transformations like sheer, translation, rotation and scaling. In contrast to this, we want to represent and control deformations that result from propagating the dynamics of objects to deformable attached objects. Our solution is to replace the linear 4x4 matrix-based transformation of a scene graph by a more generic trilinear transformation. The linear transformation allows the composition of the transformation hierarchy into one transformation. Our approach additionally allows the handling of deformations on the same level. Building on this concept we present a system capable of real-time rendering. The computations of the applied deformations of the scene graph are performed in real-time on the GPU. We allow the approximation of arbitrary nonlinear transformations and deformations by utilising grids of trilinear transformations in our system. As an application we show geometric attachments on deformable objects and their deformation on a scene graph level.

Hauser, Helwig (Ed.) et al.: Eurographics 2010. State of the Art Reports (STARs), pp. 113-136

The analysis of large graphs plays a prominent role in various fields of research and is relevant in many important application areas. Effective visual analysis of graphs requires appropriate visual presentations in combination with respective user interaction facilities and algorithmic graph analysis methods. How to design appropriate graph analysis systems depends on many factors, including the type of graph describing the data, the analytical task at hand, and the applicability of graph analysis methods. The most recent surveys of graph visualization and navigation techniques were presented by Herman et al. [HMM00] and Diaz [DPS02]. The first work surveyed the main techniques for visualization of hierarchies and graphs in general that had been introduced until 2000. The second work concentrated on graph layouts introduced until 2002. Recently, new techniques have been developed covering a broader range of graph types, such as time-varying graphs. Also, in accordance with ever growing amounts of graph-structured data becoming available, the inclusion of algorithmic graph analysis and interaction techniques becomes increasingly important. In this State-of-the-Art Report, we survey available techniques for the visual analysis of large graphs. Our review firstly considers graph visualization techniques according to the type of graphs supported. The visualization techniques form the basis for the presentation of interaction approaches suitable for visual graph exploration. As an important component of visual graph analysis, we discuss various graph algorithmic aspects useful for the different stages of the visual graph analysis process.

Lensch, Hendrik P. A. (Ed.) et al.: Eurographics 2010. Short Papers, pp. 49-52

Spectral rendering takes the full visible spectrum into account when calculating light-surface interaction and can overcome the well-known deficiencies of rendering with tristimulus color models. We present a variant of the precomputed radiance transfer algorithm that is tailored towards real-time spectral rendering on modern graphics hardware. Our method renders diffuse, self-shadowing objects with spatially varying spectral reflectance properties under distant, dynamic, full-spectral illumination. To achieve real-time frame rates and practical memory requirements we split the light transfer function into an achromatic part that varies per vertex and a wavelength-dependent part that represents a spectral albedo texture map. As an additional optimization, we project reflectance and illuminant spectra into an orthonormal basis. One area of application for our research is virtual design applications that require relighting objects with high color fidelity at interactive frame rates.

Institute for Systems and Technologies of Information, Control and Communication (INSTICC): VISIGRAPP 2010. Proceedings : International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications. INSTICC Press, 2010, pp. 165-172

Spectral rendering takes the full visible spectrum into account when calculating light-surface interaction and can overcome the well-known deficiencies of rendering with tristimulus color models. In this paper we show how to represent wavelength-dependent BRDFs as factored tensors. We use this representation for realtime spectral rendering on modern graphics hardware. Strong data compression and fast rendering times are achieved for mostly diffuse and moderately glossy isotropic surfaces. The method can handle high-resolution tabulated BRDFs, including non-reciprocal ones, which makes it well-suited for measured data. We analyze our approach numerically and visually. One area of application for our research is virtual design applications that require high color fidelity at interactive frame rates.

Institute for Systems and Technologies of Information, Control and Communication (INSTICC): VISIGRAPP 2010. Proceedings : International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications. INSTICC Press, 2010, pp. 12-19

For applications like segmentation, feature extraction and classification of point sets it is essential to know the principal curvatures and the corresponding principal directions. For the purpose of curvature estimation conformal geometric algebra promises to be a natural mathematical language: Local curvatures can be described with the help of osculating circles or spheres. On one hand, conformal geometric algebra is able to directly compute with these geometric objects, as well as with lines and planes needed for the description of vanishing curvature. On the other hand, distance measures for fitting these objects into point sets can be handled in a linear way, leading to efficient algorithms. In this paper we use conformal geometric algebra advantageously in order to locally compute continuous curvatures as well as principal curvatures of point sets without the need of costly pre-processing of raw data. We show results on artificial and real data. Numerical verification on artificial data shows the accuracy of our approach.

Institute for Systems and Technologies of Information, Control and Communication (INSTICC): VISIGRAPP 2010. Proceedings : International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications. INSTICC Press, 2010, pp. 65-71

In this work we introduce an approach for reconstructing digital 3D models from multiple perspective line drawings. One major goal is to keep the required user interaction simple and at a minimum, while making no constraints to the objects shape. Such a system provides a useful extension for digitalization of paper-based styling concepts, which today is still a time consuming process. In the presented method the line drawings are first decomposed in curves assembling a network of curves. In a second step, the positions for the endpoints of the curves are determined in 3D, using multiple sketches and a virtual camera model given by the user. Then the shapes of the 3D curves between the reconstructed 3D endpoints are inferred. This leads to a network of 3D curves, which can be used for first visual evaluations in 3D. During the whole process only little user interaction is needed, which only takes place in the pre- and post-processing phases. The approach has been applied on multiple sketches and it is shown that the approach creates plausible results within reasonable timing.

Djemal, Khalifa (Ed.) et al.: 2nd International Conference on Image Processing Theory, Tools and Applications : IPTA 2010. Los Alamitos, Calif.: IEEE Computer Society Press, 2010, pp. 343-348

A user-drawn sketch is one of the most intuitive forms of Human Computer Interaction. Users can express their intention by sketching the specific characteristics of a target object as a rough and simple black and white hand-drawn draft image. Recent advances of tablet PC and multi-touch screen technology raised increasing interest on how users might search and retrieve the desired images in databases from a simple sketched image. In this paper, we present a new approach for content based image retrieval from a query by sketchy draft images which are not in the database. Our innovation to sketch based image retrieval systems consists of three steps: (i) Image database configuration using size normalization, edge detection, and hierarchical image classification, (ii) Tensorial feature extraction of query and image data in the topology of second-order symmetric diffusion tensor fields, and (iii) Similarity measure using eigen-features between sketched query and databases to retrieve the most similar target object. Experiments are conducted to evaluate the performance of our methodology showing an efficient and mature image retrieval system.

Institute of Electrical and Electronics Engineers (IEEE): 20th International Conference on Pattern Recognition. Proceedings : ICPR 2010. Los Alamitos, Calif.: IEEE Computer Society Conference Publishing Services (CPS), 2010, pp. 3740-3743

We present a novel human action recognition system based on segmented skeletal features which are separated into several human body parts such as face, torso and limbs. Our proposed human action recognition system consists of two steps: (i) automatic skeletal feature extraction and splitting by measuring the similarity in the space of diffusion tensor fields, and (ii) multiple kernel Support Vector Machine based human action recognition. Experimental results on a set of test database show that our proposed method is very efficient and effective to recognize human actions using few parameters, independent of dimensions, shadows, and viewpoints.

Aurélio, Campilho (Ed.) et al.: Image Analysis and Recognition. Proceedings, Part I : 7th International Conference, ICIAR 2010. Berlin; Heidelberg; New York: Springer, 2010. (Lecture Notes in Computer Science (LNCS) 6111), pp. 189-199

This paper addresses a learning-based human action recognition system from multiple images based on integrating features of segmented 3D human body parts such as face, torso, and limbs. The innovation of our proposed 3D human action recognition system consists of three parts: (1) 3D reconstruction of the target object by tracking the position of a target object in a scene to voxelize the accurate 3D human model, (2) Human body model segmentation into several human body parts using ellipsoidal models in the space of second-order three dimensional diffusion tensor fields, and (3) Classification and recognition of human actions from features of the segmented human model using Multiple-Kernel based Support Vector Machine. Experimental results on a set of test volume data show that our proposed method is very efficient to visualize and recognize the human action using few parameters which are independent to partial occlusion, dimension, and viewpoint.

Djemal, Khalifa (Ed.) et al.: 2nd International Conference on Image Processing Theory, Tools and Applications : IPTA 2010. Los Alamitos, Calif.: IEEE Computer Society Press, 2010, pp. 343-348

A user-drawn sketch is one of the most intuitive forms of Human Computer Interaction. Users can express their intention by sketching the specific characteristics of a target object as a rough and simple black and white hand-drawn draft image. Recent advances of tablet PC and multi-touch screen technology raised increasing interest on how users might search and retrieve the desired images in databases from a simple sketched image. In this paper, we present a new approach for content based image retrieval from a query by sketchy draft images which are not in the database. Our innovation to sketch based image retrieval systems consists of three steps: (i) Image database configuration using size normalization, edge detection, and hierarchical image classification, (ii) Tensorial feature extraction of query and image data in the topology of second-order symmetric diffusion tensor fields, and (iii) Similarity measure using eigen-features between sketched query and databases to retrieve the most similar target object. Experiments are conducted to evaluate the performance of our methodology showing an efficient and mature image retrieval system.

Institute for Systems and Technologies of Information, Control and Communication (INSTICC): VISIGRAPP 2010. Proceedings : International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications. INSTICC Press, 2010, pp. 72-75

The segmentation of 3D target objects into coherent subregions is one of the most important issues in computer graphics as it is applied in many applications, such as medical model visualization and analysis, 3D model retrieval and recognition, skeleton extraction, and collision detection. The goal of 3D segmentation is to separate the volume or mesh data into several subregions which have similar characteristics. In this paper, we present an efficient and accurate 3D model segmentation methodology by merging and splitting the subregions in a 3D model. Our innovative 3D model segmentation system consists of two steps: i) the ellipsoidal decomposition of unorganized 3D object using properties of three dimensional second-order diffusion tensor fields, and ii) The iteratively merging and splitting of subregions of the 3D model by measuring the similarity between neighboring regions. Experimental results are conducted to evaluate the performance of our methodology using 3D models from well-known databases and 3D target objects that are reconstructed from image sequences.

Institute for Systems and Technologies of Information, Control and Communication (INSTICC): VISIGRAPP 2010. Proceedings : International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications. INSTICC Press, 2010, pp. 292-299

Sketches drawn by users are one of the most intuitive forms of Human Computer Interaction. Users can easily express their intention by sketching simple hand-drawn lines. In this paper, we consider the problem of target object detection and retrieval from a query by a sketch which is not in the database. Our novel approach consists of three steps: (1) Preprocessing to extract the skeletal features from a sketched query using size normalization, labelling, and binarization, (2) Skeletal feature extraction of query and data images in the space of diffusion tensor fields, and (3) Similarity measure using tensorial information between sketched query and database to retrieve the most similar target object in database. Experiments are conducted to evaluate the performance of our methodology, which shows to be an efficient and mature retrieval system.

ACM SIGMM: Proceedings of the ACM Multimedia 2010 International Conference : MM'10. New York: ACM, 2010, pp. 193-200

The number of available 3D models in various areas increases steadily. Effective methods to search for those 3D models by content, rather than textual annotations are crucial. For this purpose, we propose a new approach for content based 3D model retrieval by hand-drawn sketch images. This approach to retrieve visually similar mesh models from a large database consists of three major steps: (1) suggestive contour renderings from different viewpoints to compare against the user drawn sketches; (2) descriptor computation by analyzing diffusion tensor fields of suggestive contour images, or the query sketch respectively; (3) similarity measurement to retrieve the models and the most probable view-point from which a model was sketched. Our proposed sketch based 3D model retrieval system is very robust against variations of shape, pose or partial occlusion of the user drawn sketches. Experimental results are presented and indicate the effectiveness of our approach for sketch-based 3D model retrieval.

Echizen, Isao (Ed.) et al.: Sixth International Conference on Intelligent Information Hiding and Multimedia Signal Processing. Proceedings : IIH-MSP 2010. Los Alamitos, Calif.: IEEE Computer Society, 2010, pp. 329-332

Human faces are one of the most important and most popular biometric modalities to recognize individuals in a broad range of applications. So far, Local Binary Patterns have been applied to face recognition based on 2D illumination images and near infrared images, showing good robustness, discriminative ability and computational efficiency. In this paper, this method is extended to 3D face images. We investigate the influence of several parameters of this method and show improvements in the recognition rates, proving that this new method is a very promising approach for 3D face recognition.

Oravec, Milos (Ed.): Face Recognition. Sciyo, 2010, pp. 315-328

The human face is one of the most important biometric modalities for automatic authentication. Three-dimensional face recognition exploits facial surface information. In comparison to illumination based 2D face recognition, it has good robustness and high fake resistance, so that it can be used in high security areas. Nevertheless, as in other common biometric systems, potential risks of identity theft, cross matching and exposure of privacy information threaten the security of the authentication system as well as the user's privacy. As a crucial supplementary of biometrics, the template protection technique can prevent security leakages and protect privacy. In this chapter, we show security leakages in common biometric systems and give a detailed introduction on template protection techniques. Then the latest results of template protection techniques in 3D face recognition systems are presented. The recognition performances as well as the security gains are analyzed.

European Association for Computer Graphics (Eurographics): Eurographics 2009 - Annex : Tutorials, State of the Art Reports, Short Papers, Medical Prize, Education Papers, Areas Papers, EG Workshop on Natural Phenomena. Aire-la-Ville: Eurographics Association, 2009, pp. 77-80

A successful approach in triangulating point set surfaces is to apply operations, like a projection operator for advancing front algorithms, directly to Moving-Least Squares (MLS) surfaces. The MLS method naturally handles noisy input data and is especially useful for point clouds derived from real-world solids. Unfortunately, MLS is computationally extensive and complex. We present a novel projection method that does not require solving a nonlinear optimization problem as MLS does. We create a polynomial approximation of the surface similar to MLS but our method adapts the degree of the polynomial with respect to the points to be approximated. The approximated points are iteratively collected compromising connectivity information. We enhance the orientation of the local coordinate system to further improve the method. The results confirm that our method is more robust and also accelerates triangulation due to a preprocessing step that needs to be done only once per data set.

Tai, Xue-Cheng (Ed.) et al.: Scale Space and Variational Methods in Computer Vision. Proceedings. Berlin, Heidelberg, New York: Springer, 2009. (Lecture Notes in Computer Science (LNCS) 5567), pp. 844-855

In this work we describe the possible transitions for the hierarchical structure that describes an image in Gaussian scale space. Until now, this tree structure has only been used for topological segmentation. In order to perform image matching and retrieval tasks based on this structure, one needs to know which transitions are allowed when the structure is changed under influence of one control parameter. We present a list of such transitions, enabling tree edit distance operations.

IAPR Newsletter, Vol.31 (2009), 4, pp. 3-6

This article continues the Global Pattern Recognition series that focused initially on national laboratories.

Since several decades computers are obtaining a more and more prominent place in society. Increasing computational possibilities combined with expanding storage potentialities in the field of computer vision urge for methods that can perform the tasks of identifying, comparing, and classifying objects in digital data automatically. In order to be highly user-independent and trustworthy stable for cost and time saving purposes, many methods combine sophisticated knowledge of different areas of mathematics, computer science, and human perception. In this work the mathematical concept of Singularities takes a prominent role. It describes situations that are special in some sense. A simple example is given by a topographical map with height lines describing a mountain landscape. Almost all lines are simply connected, but there are some special cases: at certain positions the lines reduce to a point, for instance at a mountain top, while at other locations (at passes) they intersect themselves. Furthermore, the height lines can have sharp corners and shown maximal and minimal bending. These features involve so-called critical points, the locally special points. A second concept used in this work is the reduction of structure. Again (topographic) maps serve as an example. The larger the scales on which they are drawn, the less the details are visible, but the more the main structure is kept. This effect is known as scale space. Instead of considering the image at one particular scale, or even a set of scales, the continuum of images at all scales, the Deep Structure, is taken.

Sester, Monika (Ed.) et al.: Advances in GIScience : Proceedings of the 12th AGILE Conference. Berlin; Heidelberg; New York: Springer, 2009. (Lecture Notes in Geoinformation and Cartography (LNGC)), pp. 173-194

This paper gives an introduction to concepts for a better expert knowledge extraction for geodata harmonisation. Geodata harmonisation involves overcoming heterogeneities on the syntactic, schematic and semantic levels, but user interaction is mostly required for the semantic in-teroperability level. Consequently, the goal is to provide tools to geodomain experts which allow them to use their knowledge to describe the conceptual schemas of their domain as well as alignments to other domain's conceptual schemas in a sufficiently expressive way without becoming experts in ontology engineering. We describe an approach for this that includes both visual analysis on the basis of geospatial instances and specific interaction guidance processes. Based on this approach, a prototypical implementation of a tool called the HUMBOLDT Alignment Editor (HALE) is introduced.

Pan, Jeng-Shyang (Ed.) et al.: Fifth International Conference on Intelligent Information Hiding and Multimedia Signal Processing. Proceedings : IIH-MSP 2009. New York: IEEE, Inc., 2009, pp. 1061-1065

Privacy protection techniques are an important supplementary of biometric systems. Their main purpose is to prevent security leakages in common biometric systems and to preserve the user's privacy. However, when cryptographic functions are used in the algorithms, randomness of biometric features is strictly required from the security point of view. This randomness is hard to achieve in many feature extraction algorithms, especially for those using the local information of biometric modality. In this paper we discuss privacy protection based on a fuzzy extractor. We show that the security of the algorithm is strongly reduced when statistical properties of biometric features as well as the details of the algorithm are known. An attack exploiting feature correlation is demonstrated.

Mohamed, Kamel (Ed.) et al.: Image Analysis and Recognition : 6th International Conference, ICIAR 2009. Berlin; Heidelberg; New York: Springer, 2009. (Lecture Notes in Computer Science (LNCS) 5627)

Biometric features provide considerable usability benefits. At the same time, the inability to revoke templates and likelihood of adversaries being able to capture features raise security concerns. Recently, several template protection mechanisms have been proposed, which provide a one-way mapping of templates onto multiple pseudo-identities. While these proposed schemes make assumptions common for cryptographic algorithms, the entropy of the template data to be protected is considerably lower per bit of key material used than assumed owing to correlations arising from the biometric features. We review several template protection schemes and existing attacks followed by a correlation analysis for a selected biometric feature set and demonstrate that these correlations leave the stream cipher mechanism employed vulnerable to, among others, known plaintext-type attacks.