Liste der Fachpublikationen

Show publication details

3DHOG for Geometric Similarity Measurement and Retrieval on Digital Cultural Heritage Archives

2016

Giuseppe De Pietro (Ed.) et al.: Intelligent Interactive Multimedia Systems and Services 2016. Switzerland: Springer International Publishing, 2016. (Smart Innovation, Systems and Technologies 55), pp. 459-469

KES International Conference on Intelligent Interactive Multimedia Systems and Services (IIMSS) <9, 2016, Puerto de la Cruz, Tenerife, Spain>

With projects such as CultLab3D, 3D Digital preservation of cultural heritage will become more affordable and with this, the number of 3D-models representing scanned artefacts will dramatically increase. However, once mass digitization is possible, the subsequent bottleneck to overcome is the annotation of cultural heritage artefacts with provenance data. Current annotation tools are mostly based on textual input, eventually being able to link an artefact to documents, pictures, videos and only some tools already support 3D models. Therefore, we envisage the need to aid curators by allowing for fast, web-based, semi-automatic, 3D-centered annotation of artefacts with metadata. In this paper we give an overview of various technologies we are currently developing to address this issue. On one hand we want to store 3D models with similarity descriptors which are applicable independently of different 3D model quality levels of the same artefact. The goal is to retrieve and suggest to the curator metadata of already annotated similar artefacts for a new artefact to be annotated, so he can eventually reuse and adapt it to the current case. In addition we describe our web-based, 3D-centered annotation tool with meta- and object repositories supporting various databases and ontologies such as CIDOC-CRM.

Show publication details
El Hakimi, Wissam; Fellner, Dieter W. (Betreuer); Sakas, Georgios (Betreuer); Schipper, Jörg (Betreuer)

Accurate 3D-Reconstruction and -Navigation for High-Precision Minimal-Invasive Interventions

2016

Darmstadt, TU, Diss., 2016

The current lateral skull base surgery is largely invasive since it requires wide exposure and direct visualization of anatomical landmarks to avoid damaging critical structures. A multi-port approach aiming to reduce such invasiveness has been recently investigated. Thereby three canals are drilled from the skull surface to the surgical region of interest: the first canal for the instrument, the second for the endoscope, and the third for material removal or an additional instrument. The transition to minimal invasive approaches in the lateral skull base surgery requires sub-millimeter accuracy and high outcome predictability, which results in high requirements for the image acquisition as well as for the navigation. Computed tomography (CT) is a non-invasive imaging technique allowing the visualization of the internal patient organs. Planning optimal drill channels based on patient-specific models requires high-accurate three-dimensional (3D) CT images. This thesis focuses on the reconstruction of high quality CT volumes. Therefore, two conventional imaging systems are investigated: spiral CT scanners and C-arm cone-beam CT (CBCT) systems. Spiral CT scanners acquire volumes with typically anisotropic resolution, i.e. the voxel spacing in the slice-selection-direction is larger than the in-the-plane spacing. A new super-resolution reconstruction approach is proposed to recover images with high isotropic resolution from two orthogonal low-resolution CT volumes. C-arm CBCT systems offers CT-like 3D imaging capabilities while being appropriate for interventional suites. A main drawback of these systems is the commonly encountered CT artifacts due to several limitations in the imaging system, such as the mechanical inaccuracies. This thesis contributes new methods to enhance the CBCT reconstruction quality by addressing two main reconstruction artifacts: the misalignment artifacts caused by mechanical inaccuracies, and the metal-artifacts caused by the presence of metal objects in the scanned region. CBCT scanners are appropriate for intra-operative image-guided navigation. For instance, they can be used to control the drill process based on intra-operatively acquired 2D fluoroscopic images. For a successful navigation, accurate estimate of C-arm pose relative to the patient anatomy and the associated surgical plan is required. A new algorithm has been developed to fulfill this task with high-precision. The performance of the introduced methods is demonstrated on simulated and real data.

Show publication details

Adaptive Semantics Visualization

2016

Studies in Computational Intelligence 646

Zugl.: Darmstadt, TU, Diss., 2014

This book introduces a novel approach for intelligent visualizations that adapts the different visual variables and data processing to human's behavior and given tasks. Thereby a number of new algorithms and methods are introduced to satisfy the human need of information and knowledge and enable a usable and attractive way of information acquisition. Each method and algorithm is illustrated in a replicable way to enable the reproduction of the entire "SemaVis" system or parts of it. The introduced evaluation is scientifically well-designed and performed with more than enough participants to validate the benefits of the methods. Beside the introduced new approaches and algorithms, readers may find a sophisticated literature review in Information Visualization and Visual Analytics, Semantics and information extraction, and intelligent and adaptive systems. This book is based on an awarded and distinguished doctoral thesis in computer science.

Show publication details

Adaptive UW Image Deblurring via Sparse Representation

2016

Santos, Luis Paulo (Ed.) et al.: Eurographics 2016. Short Papers. The Eurographics Association, 2016, pp. 41-44

Annual Conference of the European Association for Computer Graphics (Eurographics) <37, 2016, Lisbon, Portugal>

We present an adaptive underwater (UW) image deblurring algorithm based on sparse representation where a blur estimation is used to guide the algorithm for the best image reconstruction. The strong blur in this medium is caused by forward scatter and is challenging since it increases by camera scene distance. It is a common practice to use methods such as dark channel prior to estimate the depth map, and use this information to improve the image quality. However, we found it not successful in the case of blur since these methods are based on haze phenomenon. We propose a simple but effective algorithm via sparse representation which establishes a blur strength estimation and uses this information for adaptive deblurring. Extensive experiments manifest the effectiveness of our method in case of small but challenging blur changes.

Show publication details

Adaptives und hybrides SLAM für handgeführte RGBD-Kameras

2016

Darmstadt, TU, Master Thesis, 2016

Mit der steigenden Beliebtheit von RGBD-Sensoren wurde viel Forschung im Bereich der Aufnahme und Rekonstruktion von dreidimensionalen Umgebungen mit Hilfe von solchen Sensoren betrieben. Für die Konstruktion muss das sogenannte Simultaneous Localization and Mapping (SLAM)-Problem gelöst werden. Die meisten RGBD-SLAM-Systeme verwenden hierbei den punktbasierten Iterative Closest Point (ICP)-Algorithmus. Auch wenn ICP ein gut untersuchter Algorithmus ist, so stößt er bei verrauschten Daten und besonders bei texturarmen Bereichen mit wenigen geometrischen Merkmalen, wie z.B. großen leeren Flächen, auf Probleme. Eine Option, diese Limitierung anzugehen, ist das zusätzliche Ausnutzen von Ebenen in der Szene, besonders da sie die häufigste Form in von Menschen erbauten Innenräumen und Außenanlagen sind. Taguchi et al. [TJRF13] veröffentlichte 2013 die erste globale Registrierungsmethode, in welcher Punkt-zu-Punkt- und Ebene-zu-Ebene-Korrespondenzen zu einem echtzeitfähigen SLAM-System vereint werden. Kurz darauf folgte die Publikation von Ataer-Cansizoglu et al. [ACTRG13], welche zusätzlich ein Bewegungsvorhersage-Modell ausnutzt, um Korrespondenzen zu bestimmen. Ein Nachteil dieser Verfahren ist die hohe Verarbeitungszeit eines Registrierungsschrittes. Dieser bewirkt, dass die Verfahren nicht in der Lage sind, interaktive Rekonstruktionen durchzuführen. Das Ziel dieser Arbeit ist die Implementierung eines SLAM-Algorithmus für handgeführte RGBDKameras, der sowohl Punkte, als auch Flächen zur Registrierung nutzt. Im Gegensatz zu bestehenden Verfahren wird in dieser Arbeit ein lokaler Registrierungsalgorithmus umgesetzt. Flächenmerkmale werden bevorzugt verwendet, da ihre Anzahl in Szenen signifikant geringer ist als die von Punkten. Das ermöglicht eine schnellere Korrespondenzsuche und Registrierung. Dem zugrundeliegenden RANSACbasierten Algorithmus reicht bereits eine minimale Anzahl an Korrespondenzen aus, um die Sensorpose zu bestimmen. Somit ist der Algorithmus in der Lage, die Registrierung auch in texturarmen Bereichen mit wenigen geometrischen Merkmalen durchzuführen, in denen Techniken, welche nur Punkte benutzen, scheitern. Des Weiteren ermöglicht der lokale Registrierungsansatz eine interaktive Nutzung, um dem Nutzer in Echtzeit Rückmeldung über den Registrierungsprozess zu geben. Zusätzlich implementierte Erweiterungen, welche die detektierten Flächeninformationen zur Geometriekorrektur ausnutzen, unterstützen den Registrierungsvorgang. Durchgeführte Experimente demonstrieren eine interaktive Rekonstruktion von Innenräumen mit einer handgeführten RGBD-Kamera, einer Kinect. Zudem weist das System im Gegensatz zu vergleichbaren hybriden Systemen eine sechsfach höhere Rekonstruktionsrate auf. Bei der Gegenüberstellung anhand eines Benchmark-Datensatzes für RGBD-Sensoren konnte des Weiteren in texturarmen Umgebungen eine Überlegenheit gegenüber punktbasierten Verfahren nachgewiesen werden.

Show publication details
Weinmann, Michael; Langguth, Fabian; Goesele, Michael; Klein, Reinhard

Advances in Geometry and Reflectance Acquisition

2016

Sousa, Augusto (Ed.) et al.: Eurographics 2016. Tutorials. The Eurographics Association, 2016, 71 p.

Annual Conference of the European Association for Computer Graphics (Eurographics) <37, 2016, Lisbon, Portugal>

This tutorial is focused on acquisition methods for geometry and reflectance as well as strategies towards an efficient acquisition pipeline to fulfill the demands of industry with respect to mass digitization of 3D contents. We provide a thorough overview of the standard methods for the acquisition of both geometry and reflectance of surfaces with different types of reflectance behavior ranging from diffuse over opaque to specular surfaces or even translucent and transparent surfaces as well as the necessary preliminaries of material appearance and setup calibration. As standard acquisition techniques are only well-suited for a limited range of surface materials, we will also discuss strategies on how an efficient, fully automatic acquisition can still be achieved when no prior information with respect to the surface reflectance behavior is available. In addition, a discussion of strategies regarding an acquisition in the wild, i.e. under uncontrolled conditions, is provided.

Show publication details
Matthies, Denys J.C.; Bieber, Gerald; Kaulbars, Uwe

AGIS: Automated Tool Detection & Hand-Arm Vibration Estimation using an Unmodified Smartwatch

2016

Matthies, Denys J.C. (Ed.) et al.: iWOAR 2016 : 3rd international Workshop on Sensor-based Activity Recognition and Interaction. New York: ACM Press, 2016. (ACM International Conference Proceedings Series 1183), Art. No. 8, 4 p.

International Workshop on Sensor-based Activity Recognition (iWOAR) <3, 2016, Rostock, Germany>

Over the past three decades, it has been known that longlasting and intense hand-arm vibrations (HAV) can cause serious diseases, such as the Raynaud- / White Finger- Syndrome. In order to protect workers nowadays, the longterm use of tools such as a drill, grinder, rotary hammer etc. underlie strict legal regulations. However, users rarely comply with these regulations because it is quite hard to manually estimate vibration intensity throughout the day. Therefore, we propose a wearable system that automatically counts the daily HAV exposure doses due to the fact that we are able to determine the currently used tool. With the implementation of AGIS, we demonstrate the technical feasibility of using the integrated microphone and accelerometer from a commercial smartwatch. In contrast to prior works, our approach does not require a technical modification of the smartwatch nor an instrumentation of the environment or the tool. A pilot study shows our proofof- concept to be applicable in real workshop environments.

Show publication details

Ambient Assisted Living

2016

Fischer, Florian (Ed.) et al.: eHealth in Deutschland : Anforderungen und Potenziale innovativer Versorgungsstrukturen. Springer, 2016, pp. 203-222

Das Anwendungsfeld Ambient Assisted Living (AAL) beschreibt technische Systeme zur Unterstützung hilfsbedürftiger Personen im Alltag. In den vergangenen Jahren wurde in Deutschland und Europa viel in die Entwicklung und Erprobung von Technologien zur Unterstützung in der häuslichen Umgebung investiert, jedoch häufig ohne nachhaltige Effekte am Markt. Ein fehlender Aspekt war häufig die mangelnde Involvierung aller notwendigen Parteien. In diesem Kapitel werden die Potenziale assistiver Technologien beleuchtet, eine Studie zur Akzeptanz derartiger Technologien bei Senioren vorgestellt sowie ein Ausblick auf zukünftige Entwicklungen in diesem Bereich präsentiert.

Show publication details
Fehling, Christian Dominic; Goertz, Lutz; Hagenhofer, Thomas; Müller, Andreas

Ausbildungsqualität 4.0 - ein Blick in die Zukunft am Beispiel des Projektes Social Augmented Learning

2016

Schröder, Frank (Ed.): Unternehmensstrategie Ausbildungsqualität : Berliner Initiativen für kleine und mittlere Unternehmen in einem sich wandelnden Arbeitsmarkt. Bielefeld: wbv, 2016, pp. 155-162

Die Transformation hin zu einer Industrie 4.0, getragen durch einen bereits im Mainstream wahrgenommenen, ausgesprochen technologischen Entwicklungsschub stellt zahlreiche Berufsfelder vor nicht zu unterschätzende Herausforderungen. Das Berufsfeld Medientechnologin bzw. Medientechnologe Druck, das in diesem Beitrag und bei den Arbeiten im Projekt "Social Augmented Learning" im Mittelpunkt steht, ist davon ebenso stark betroffen wie andere Berufsfelder mit vergleichsweise hohem Anteil an Routinetätigkeiten im Kontext von Automatisierung und der Arbeit an, mit und in Bezug zu Maschinen. Ziel des Projektes war und ist es, nicht nur die technische Infrastruktur für ein effektives und effizientes digitales Lernen mit mobilen Endgeräten und mit Augmented Reality zu ermöglichen, zu definieren und zu beschreiben, sondern zudem in eine kohärente Medienstrategie für Berufsschulen und Ausbildungsbetriebe zu überführen.

Show publication details
Hiemenz, Benedikt; Kuijper, Arjan (Betreuer); Krämer, Michel (Betreuer)

Authentication and Searchable Symmetric Encryption for Cloud-Based Storage of Geospatial Data

2016

Darmstadt, TU, Master Thesis, 2016

Many companies outsource their data and computations to the cloud to benefit from a worldwide data access and economic advantages such as a flexible storage management. Since cloud infrastructures are usually provided by third parties, companies must protect their private data before moving them. An example are geospatial data sets storing confidential information about urban areas. To secure these data, we present two security extensions for cloud-based geospatial data storages. First, we introduce an authentication process enabling applications to securely identify their users. We build our approach on token-based techniques but enhance them by essential features such as a revocation mechanism allowing administrators to block certain users at any time. Our extension requires minimal information on the server side and is suitable for distributed environments like a cloud infrastructure. Our main security extension allows users to encrypt their geospatial data and make them searchable at the same time. We develop a Searchable Symmetric Encryption protocol using an index to provide search features on encrypted data. Our approach is dynamic as we do not require an initialization phase and allow users to constantly add new data to the index or remove existing records. In the course of this thesis, we design multiple versions of our protocol differing in their level of security and performance respectively. All of them support boolean expressions as part of the query and geospatial-related search criteria such as bounding boxes. On the client side, our protocol facilitates multi-device support because we only expect cryptographic keys. Our findings indicate that both extensions are suitable for real-world applications in distributed systems but the search in encrypted data requires some more runtime than in plaintext data.

Show publication details
Noll, Matthias; Nadolny, Anne; Wesarg, Stefan

Automated Kidney Detection for 3D Ultrasound Using Scan Line Searching

2016

The International Society for Optical Engineering (SPIE): Proceedings of SPIE, Article 9790-46, 6 p.

SPIE Medical Imaging Symposium <2016, San Diego, CA, USA>

Ultrasound (U/S) is a fast and non-expensive imaging modality that is used for the examination of various anatomical structures, e.g. the kidneys. One important task for automatic organ tracking or computer-aided diagnosis is the identification of the organ region. During this process the exact information about the transducer location and orientation is usually unavailable. This renders the implementation of such automatic methods exceedingly challenging. In this work we like to introduce a new automatic method for the detection of the kidney in 3D U/S images. This novel technique analyses the U/S image data along virtual scan lines. Here, characteristic texture changes when entering and leaving the symmetric tissue regions of the renal cortex are searched for. A subsequent feature accumulation along a second scan direction produces a 2D heat map of renal cortex candidates, from which the kidney location is extracted in two steps. First, the strongest candidate as well as its counterpart are extracted by heat map intensity ranking and renal cortex size analysis. This process exploits the heat map gap caused by the renal pelvis region. Substituting the renal pelvis detection with this combined cortex tissue feature increases the detection robustness. In contrast to model based methods that generate characteristic pattern matches, our method is simpler and therefore faster. An evaluation performed on 61 3D U/S data sets showed, that in 55 cases showing none or minor shadowing the kidney location could be correctly identified.

Show publication details
Bartschat, Andreas; Kuijper, Arjan (Betreuer)

Automatic Classification of Cornea Tissues for Autofocus Algorithm

2016

Darmstadt, TU, Master Thesis, 2016

Corneal confocal microscopy (CCM) is a spreading technique for investigations of cellular structures in the human cornea. It is non-invasive and allows in vivo imaging of the different tissue layers in the cornea with high resolution. High expectations are currently placed on CCM, to allow rapid and detailed analysis of pathological alterations affecting the peripheral nerves that innervate the cornea, resulting not only in fast diagnosis, but also providing insights into the progress and severity of diseases like diabetes. For the fast and reliable imaging of the sub-basal nerve plexus (SNP), the layer with the highest density of nerves, the focus must be adapted to compensate anatomical layer irregularities and reversible folds. This thesis analyses classification methods of the anatomical tissues surrounding the SNP, to find reliable and fast classification models, suitable for online focus adaptations of the microscope to the layer of interest based on the predicted tissue type of the current image. The proposed methods are able to achieve accuracies of more than 92% with a runtime of less than 10 ms per image, evaluated on datasets of more than 8000 images as well as successfully applied in studies for the reconstruction of large field of view images of the SNP.

Show publication details
Bartschat, Andreas; Toso, Lorenzo; Stegmaier, Johannes; Kuijper, Arjan; Mikut, Ralf; Köhler, Bernd; Allgeier, Stephan

Automatic Corneal Tissue Classification Using Bag-Of-Visual-Words Approaches

2016

Heizmann, M. (Ed.) et al.: Forum Bildverarbeitung 2016. Karlsruhe: KIT Scientific Publishing, 2016, pp. 245-256

Forum Bildverarbeitung <1, 2016, Karlsruhe, Deutschland>

Corneal confocal microscopy is a promising diagnostic method for peripheral neuropathy. It allows the recording of the sub-basal nerve plexus (SNP) and enables the morphological analysis of peripheral nerves. This work evaluates classification models for real-time evaluation of cornea images in order to find suitable methods for an automatic focus adaptation to the SNP. The analyzed Bag-of-Visual-Words method leads to models with an accuracy of 0.9, even on a small training dataset, and a runtime of 18 ms per image. Furthermore, the analysis of the support vector machine real-valued output shows high potential for the implementation of real-time focus optimization methods.

Show publication details

An Automatic Free Fluid Detection for Morrison´s-Pouch

2016

Shekhar, Raj (Ed.) et al.: Clinical Image-Based Procedures. Translational Research in Medical Imaging : 5th International Workshop, CLIP 2016. Held in Conjunction with MICCAI 2016. Berlin, Heidelberg, New York: Springer, 2016. (Lecture Notes in Computer Science (LNCS) 9958), pp. 77-84

International Workshop on Clinical Image-based Procedures (CLIP) <5, 2016, Athens, Greece>

Ultrasound provides a useful and readily available imaging tool to detect free fluids in blunt abdominal trauma patients. However, applying conventional 2D ultrasound to diagnose the patient requires a well trained physician. In this paper we describe a fully automatic free fluid detection pipeline for the hepathorenal recess or Morrison's pouch using 3D ultrasound acquisitions. The image data is collected using the standardized "Focused Assessment with Sonography for Trauma" (FAST) exam. Our method extracts key structures like the kidney and the liver from the image data and uses their relative positions to search and detect free fluids between the organ interfaces. To evaluate our method we have developed a free fluid simulation that allows us to generate free fluid images using acquisitions of healthy volunteers. Our intentions are to enable even untrained ultrasound operators to perform a free fluid diagnosis of an injured person. In order to do this, our method additionally provides basic image acquisition guidance information.

Show publication details
Kutlucan, Osman; Kuijper, Arjan (Betreuer); Kirchbuchner, Florian (Betreuer)

Barrierefreies Lagersystem zur Unterstützung von Menschen mit eingeschränkter visueller Wahrnehmungsfähigkeit

2016

Darmstadt, TU, Master Thesis, 2016

Im Mittelpunkt dieser Masterarbeit steht die Realisierung eines Lagersystems für Menschen mit eingeschränkter visueller Wahrnehmungsfähigkeit. Das in diesem Rahmen entwickelte Lagersystem verwendet Hand- und Gestenerkennung, Spracherkennung, Sprachsynthese und Sonifikation. Das Ziel des Systems ist es, sowohl eine pervasive Benutzerschnittstelle anzubieten, welche es blinden Benutzern ermöglicht, mit dem Lagerbereich auf eine natürliche Weise zu interagieren, und ein Lagersystem zu haben, welches mit geringem Aufwand konfiguriert werden kann. Daher werden in der Thesis in erster Linie verschiedene verwandte Ansätze und Konzepte betrachtet, woraufhin beschrieben wird, wie das vorgeschlagene Konzept der Thesis entwickelt wurde und warum dieses Konzept sich besser für blinde Benutzer eignet. Um dies zu analysieren, wird das vorgeschlagene System iterativ evaluiert. Im Rahmen der Evaluation wurden Verbesserungsvorschläge aufgenommen, und die Präzision und Effizienz der Implementierung gemessen. Die Ergebnisse der Evaluation dieser Thesis zeigen, dass das Konzept des vorgeschlagenen Systems angemessen ist und von blinden Benutzern gut aufgenommen wird, was z.B. durch einen Teilnehmer der Evaluation bei der Beurteilung eines Subworkflows des Systems mit folgenden Worten bestätigt wurde: "Die Interaktion fühlt sich so an, als wäre man nicht blind". Aber die Implementierung des Systems weist einige Probleme auf, welche sich vorwiegend in der Hand- und Gestenerkennung des Systems zeigen. Mit Behebung dieser Probleme könnte die User Experience einen höheren Grad erreichen, wodurch das System im alltäglichen Leben einsetzbar werden könnte.

Show publication details

Benchmarking Sensors in Smart Environments - Method and Use Cases

2016

Journal of Ambient Intelligence and Smart Environments, Vol.8 (2016), 6, pp. 645-664

Smart environment applications can be based on a large variety of different sensors that may support the same use case, but have specific advantages or disadvantages. Benchmarking can allow determining the most suitable sensor systems for a given application by calculating a single benchmarking score, based on weighted evaluation of features that are relevant in smart environments. This set of features has to represent the complexity of applications in smart environments. In this work we present a benchmarking model that can calculate a benchmarking score, based on nine selected features that cover aspects of performance, the environment and the pervasiveness of the application. Extensions are presented that normalize the benchmark-ing score if required and compensate central tendency bias, if necessary. We outline how this model is applied to capacitive proximity sensors that measure properties of conductive objects over a distance. The model is used to identify existing and find potential new application domains for this upcoming technology in smart environments.

Show publication details
Schinko, Christoph; Peer, Markus; Hammer, Daniel; Pirstinger, Matthias; Lex, Cornelia; Koglbauer, Ioana; Eichberger, Arno; Holzinger, Jürgen; Eggeling, Eva; Fellner, Dieter W.; Ullrich, Torsten

Building a Driving Simulator with Parallax Barrier Displays

2016

Magnenat-Thalmann, Nadia (Ed.) et al.: Proceedings of the 11th Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications. Volume 1 : 11th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications. SciTePress, 2016, pp. 283-291

International Joint Conference on Computer Vision and Computer Graphics Theory and Applications (VISIGRAPP) <11, 2016, Rome, Italy>

In this paper, we present an optimized 3D stereoscopic display based on parallax barriers for a driving simulator. The overall purpose of the simulator is to enable user studies in a reproducible environment under controlled conditions to test and evaluate advanced driver assistance systems. Our contribution and the focus of this article is a visualization based on parallax barriers with (I) a-priori optimized barrier patterns and (II) an iterative calibration algorithm to further reduce visualization errors introduced by production inaccuracies. The result is an optimized 3D stereoscopic display perfectly integrated into its environment such that a single user in the simulator environment sees a stereoscopic image without having to wear specialized eye-wear.

Show publication details

c-Space: Time-evolving 3D Models (4D) from Heterogeneous Distributed Video Sources

2016

Catalano, Chiara Eva (Ed.) et al.: GCH 2016 : Eurographics Workshop on Graphics and Cultural Heritage. Goslar: Eurographics Association, 2016, pp. 12-18

Eurographics Symposium on Graphics and Cultural Heritage (GCH) <14, 2016, Genova, Italy>

We introduce c-Space, an approach to automated 4D reconstruction of dynamic real world scenes, represented as time-evolving 3D geometry streams, available to everyone. Our novel technique solves the problem of fusing all sources, asynchronously captured from multiple heterogeneous mobile devices around a dynamic scene at a real word location. To this end all captured input is broken down into a massive unordered frame set, sorting the frames along a common time axis, and finally discretizing the ordered frame set into a time-sequence of frame subsets, each subject to photogrammetric 3D reconstruction. The result is a time line of 3D models, each representing a snapshot of the scene evolution in 3D at a specific point in time. Just like a movie is a concatenation of time-discrete frames, representing the evolution of a scene in 2D, the 4D frames reconstructed by c-Space line up to form the captured and dynamically changing 3D geometry of an event over time, thus enabling the user to interact with it in the very same way as with a static 3D model. We do image analysis to automatically maximize the quality of results in the presence of challenging, heterogeneous and asynchronous input sources exhibiting a wide quality spectrum. In addition we show how this technique can be integrated as a 4D reconstruction web service module, available to mobile end-users.

Show publication details
Sheldrick, Peter; Wuest, Harald (Betreuer); Kuijper, Arjan (Betreuer)

CAD-Model Tracking using RGB-D Cameras

2016

Darmstadt, TU, Master Thesis, 2016

This thesis deals with the determination of the six DOF of an RGB-D camera relative to a known CAD-Model. Extracting features in image based tracking with no other input data reduces the achievable precision of tracking. This thesis presents methods that use the whole input frame from a depth camera - these are so called "dense'" methods. Methods such as ICP, that is used in KinectFusion, and depth image warping, which is used in DVO-SLAM, are compared for the task of CAD-Model tracking. Rendering is used for tracking and both GPU implementations such as OpenGL and CPU ray casting is used to track real depth data.

Show publication details

Calibration of Shared Flat Refractive Stereo Systems

2016

Aurélio, Campilho (Ed.) et al.: Image Analysis and Recognition : International Conference. Switzerland: Springer International Publishing, 2016. (Lecture Notes in Computer Science (LNCS) 9730), pp. 433-442

International Conference on Image Analysis and Recognition (ICIAR) <13, 2016, Póvoa de Varzim, Portugal>

The calibration of underwater camera systems differs significantly from calibration in air due to the refraction of light. In this paper, we present a calibration approach for a shared flat refractive stereo system that is based on virtual object points. We propose a sampling strategy in combination with an efficiently solvable set of equations for the calibration of the refractive parameters. Due to the independence of calibration targets of known dimensions, the approach can be realized by using stereo correspondences alone.

Show publication details
Braun, Andreas; Zander-Walz, Sebastian; Krepp, Stefan; Rus, Silvia; Wichert, Reiner; Kuijper, Arjan

CapTap - Combining Capacitive Gesture Recognition and Acoustic Touch Detection

2016

Matthies, Denys J.C. (Ed.) et al.: iWOAR 2016 : 3rd international Workshop on Sensor-based Activity Recognition and Interaction. New York: ACM Press, 2016. (ACM International Conference Proceedings Series 1183), Art. No. 6, 6 p.

International Workshop on Sensor-based Activity Recognition (iWOAR) <3, 2016, Rostock, Germany>

Capacitive sensing is a common technology for finger-controlled touch screens. The variety of proximity sensors extends the range, thus supporting mid-air gesture interaction and application below any non-conductive materials. However, this comes at the cost of limited resolution for touch detection. In this paper, we present CapTap, which uses capacitive proximity and acoustic sensing to create an interactive surface that combines mid-air and touch gestures, while being invisibly integrated into living room furniture. We introduce capacitive imaging, investigating the use of computer vision methods to track hand and arm positions and present several use cases for CapTap. In a user study we found that the system has average localization errors of 1.5cm at touch distance and 5cm at an elevation of 20cm above the table. The users found the system intuitive and interesting to use.

Show publication details
Cordts, Marius; Omran, Mohamed; Ramos, Sebastian; Rehfeld, Timo; Enzweiler, Markus; Benenson, Rodrigo; Franke, Uwe; Roth, Stefan; Schiele, Bernt

The Cityscapes Dataset for Semantic Urban Scene Understanding

2016

IEEE Computer Society: 29th IEEE Conference on Computer Vision and Pattern Recognition. Proceedings : CVPR 2016. IEEE Computer Society Conference Publishing Services (CPS), 2016, pp. 3213-3223

IEEE Conference on Computer Vision and Pattern Recognition (CVPR) <2016, Las Vegas, Nevada, USA>

Visual understanding of complex urban street scenes is an enabling factor for a wide range of applications. Object detection has benefited enormously from large-scale datasets, especially in the context of deep learning. For semantic urban scene understanding, however, no current dataset adequately captures the complexity of real-world urban scenes. To address this, we introduce Cityscapes, a benchmark suite and large-scale dataset to train and test approaches for pixel-level and instance-level semantic labeling. Cityscapes is comprised of a large, diverse set of stereo video sequences recorded in streets from 50 different cities. 5000 of these images have high quality pixel-level annotations; 20 000 additional images have coarse annotations to enable methods that leverage large volumes of weakly-labeled data. Crucially, our effort exceeds previous attempts in terms of dataset size, annotation richness, scene variability, and complexity. Our accompanying empirical study provides an in-depth analysis of the dataset characteristics, as well as a performance evaluation of several state-of-the-art approaches based on our benchmark.

Show publication details
Shekhar, Raj; Wesarg, Stefan; González Ballester, Miguel Angel; Drechsler, Klaus; Sato, Yoshinobu; Erdt, Marius; Linguraru, Marius George; Oyarzun Laura, Cristina

Clinical Image-Based Procedures. Translational Research in Medical Imaging: 5th International Workshop, CLIP 2016. Held in Conjunction with MICCAI 2016

2016

International Workshop on Clinical Image-based Procedures (CLIP) <5, 2016, Athens, Greece>

Lecture Notes in Computer Science (LNCS) 9958

Show publication details

Combining Low-level Features of Offline Questionnaires for Handwriting Identification

2016

Aurélio, Campilho (Ed.) et al.: Image Analysis and Recognition : International Conference. Switzerland: Springer International Publishing, 2016. (Lecture Notes in Computer Science (LNCS) 9730), pp. 46-54

International Conference on Image Analysis and Recognition (ICIAR) <13, 2016, Póvoa de Varzim, Portugal>

When using anonymous offline questionnaires for reviewing services or products it is often not guaranteed that a reviewer does this only once as intended. In this paper an applied combination of different features of handwritten characteristics and its fusion is presented to expose such manipulations. The presented approach covers the aspects of alignment normalization, segmentation, feature extraction, classification and fusion. Nine features from handwritten text, numbers and checkboxes are extracted and used to recognize handwriter duplicates. The proposed method has been tested on a novel database containing pages of handwritten text produced by 1,734 writers. Furthermore we show that the unified biometric decision using a weighted sum combination rule can significantly improve writer identification performance even on low level features.

Show publication details
Landesberger, Tatiana von; Basgier, Dennis; Becker, Meike

Comparative Local Quality Assessment of 3D Medical Image Segmentations with Focus on Statistical Shape Model-based Algorithms

2016

IEEE Transactions on Visualization and Computer Graphics, Vol.22 (2016), 12, pp. 2537-2549. Published online: 19 November 2015

The quality of automatic 3D medical segmentation algorithms needs to be assessed on test datasets comprising several 3D images (i.e., instances of an organ). The experts need to compare the segmentation quality across the dataset in order to detect systematic segmentation problems. However, such comparative evaluation is not supported well by current methods. We present a novel system for assessing and comparing segmentation quality in a dataset with multiple 3D images. The data is analyzed and visualized in several views. We detect and show regions with systematic segmentation quality characteristics. For this purpose, we extended a hierarchical clustering algorithm with a connectivity criterion. We combine quality values across the dataset for determining regions with characteristic segmentation quality across instances. Using our system, the experts can also identify 3D segmentations with extraordinary quality characteristics. While we focus on algorithms based on statistical shape models, our approach can also be applied to cases, where landmark correspondences among instances can be established. We applied our approach to three real datasets: liver, cochlea and facial nerve. The segmentation experts were able to identify organ regions with systematic segmentation characteristics as well as to detect outlier instances.