Die »Selected Readings in Computer Graphics 2017« bestehen aus 40 ausgewählten Artikeln von insgesamt 133 wissenschaftlichen Veröffentlichungen.
Die Beiträge kommen aus dem Fraunhofer-Institut für Graphische Datenverarbeitung IGD mit Standorten in Darmstadt wie auch in Rostock, Singapur und Graz, den Partner-Instituten an den jeweiligen Universitäten, der Fachgruppe Graphisch-Interaktive Systeme der Technischen Universität Darmstadt, der Computergraphics and Communication Gruppe am Institut für Informatik der Universität Rostock, der Nanyang Technological University (NTU), Singapur, und dem Visual Computing Excellenz-Cluster der Technischen Universität Graz. Sie alle arbeiten eng in Projekten sowie Forschung und Entwicklung im Gebiet der Computer Graphik zusammen.
Alle Artikel erschienen vorher in verschiedenen wissenschaftlichen Büchern, Zeitschriften, Konferenzbänden und Workshops. Die Veröffentlichungen mussten einen gründlichen Begutachtungsprozess durch international führende Experten und etabilierte technische Vereinigungen durchlaufen. Deshalb geben die Selected Readings einen recht guten und detaillierten Überblick über die wissenschaftlichen Entwicklungen in der Computer Graphik im Jahr 2017. Sie werden von Professor Dieter W. Fellner, dem Leiter des Fraunhofer-Instituts für Graphische Datenverarbeitung IGD in Darmstadt zusammengestellt. Er ist zugleich Professor am Fachbereich Informatik der Technischen Universität Darmstadt und Professor an der Fakultät für Informatik der Technischen Universität Graz.
Die Selected Readings in Computer Graphics 2017 befassen sich mit Aspekten und Trends der Forschung und Entwicklung in Computer Graphik auf den Gebieten
Liste der Publikationen
3D Mass Digitization: A Milestone for Archeological Documentation
VAR. Virtual Archaeology Review [online]
In the heritage field the demand for fast and efficient 3D digitization technologies for historic remains is increasing. Besides, 3D digitization has proved to be a promising approach to enable precise reconstructions of objects. Yet, unlike the digital acquisition of cultural goods in 2D widely used today, 3D digitization often still requires a significant investment of time and money. To make it more widely available to heritage institutions, the Competence Center for Cultural Heritage Digitization at the Fraunhofer Institute for Computer Graphics Research IGD has developed CultLab3D, the world's first fully automatic 3D mass digitization facility for collections of three-dimensional objects. CultLab3D is specifically designed to automate the entire 3D digitization process thus allowing users to scan and archive objects on a large-scale. Moreover, scanning and lighting technologies are combined to capture the exact geometry, texture, and optical material properties of artefacts to produce highly accurate photo-realistic representations. The unique setup allows shortening the time needed for digitization to several minutes per artefact instead of hours, as required by conventional 3D scanning methods.
3D Meta Model Generation with Application in 3D Object Retrieval
CGI 2017. Proceedings of the Computer Graphics International Conference
Computer Graphics International (CGI) <34, 2017, Yokohama, Japan>
In the application of 3D object retrieval we search for 3D objects similar to a given query object. When a user searches for a certain class of objects like 'planes' the results can be unsatisfying: Many object variations are possible for a single class and not all of them are covered with one or a few example objects. We propose a meta model representation which corresponds to a procedural model with meta-parameters. Changing the meta-parameters leads to different variations of a 3D object. For the meta model generation a single object is constructed with a modeling tool. We automatically extract a procedural representation of the object. By inserting metaparameters we generate our meta model. The meta model defines a whole object class. The user can choose a meta model and search for all objects similar to any instance of the meta model to retrieve all objects of a certain class from a 3D object database. We show that the retrieval precision is signifcantly improved using the meta model as retrieval query.
A Data-Driven Point Cloud Simplification Framework for City-Scale Image-Based Localization
IEEE Transactions on Image Processing
City-scale 3D point clouds reconstructed via structure-from-motion from a large collection of Internet images are widely used in the image-based localization task to estimate a 6-DOF camera pose of a query image. Due to prohibitive memory footprint of city-scale point clouds, image-based localization is difficult to be implemented on devices with limited memory resources. Point cloud simplification aims to select a subset of points to achieve a comparable localization performance using the original point cloud. In this paper, we propose a data-driven point cloud simplification framework by taking it as a weighted K-Cover problem, which mainly includes two complementary parts. First, a utility-based parameter determination method is proposed to select a reasonable parameter K for K-Cover-based approaches by evaluating the potential of a point cloud for establishing sufficient 2D-3D feature correspondences. Second, we formulate the 3D point cloud simplification problem as a weighted K-Cover problem, and propose an adaptive exponential weight function based on the visibility probability of 3D points. The experimental results on three popular datasets demonstrate that the proposed point cloud simplification framework outperforms the state-of-the-art methods for the image-based localization application with a well predicted parameter in the K-Cover problem.
Accurate Physics-Based Registration for the Outcome Validation of Minimal Invasive Interventions and Open Liver Surgeries
IEEE Transactions on Biomedical Engineering
The purpose of this paper is to present an outcome validation tool for tumor radiofrequency (RF) ablation and resection. Methods: Intervention assessment tools require an accurate registration of both pre- and postoperative computed tomographies able to handle big deformations. Therefore, a physics-based method is proposed with that purpose. To increase the accuracy both automatically detected internal and surface physical landmarks are incorporated in the registration process. Results: The algorithm has been evaluated in 25 clinical datasets containing RF ablations, resections, and patients with recurrent tumors. The achieved accuracy is 1.2 mm measured as mean internal distance between vessel landmarks and a positive predictive value of 0.95. The quantitative and qualitative results of the outcome validation tool show that in 50% of the cases tumors were only partially covered by the treatment. Conclusion: The use of internal and surface landmarks combined with a physics-based registration method increases the accuracy of the results compared to the accuracy of state of the art methods. An accurate outcome validation tool is important in order to certify that the tumor and its safety margin were fully covered by the treatment. Significance: An accurate outcome validation tool can result in a decrease of the tumor recurrence rate.
An Exploratory Study on Electric Field Sensing
European Conference on Ambient Intelligence (AmI) <13, 2017, Malaga, Spain>
Electric fields are influenced by the human body and other conducting materials. Capacitive measurement techniques are used in touch-screens, in the automobile industry, and for presence and activity recognition in Ubiquitous Computing. However, a drawback of the capacitive technology is the energy consumption, which is an important aspect for mobile devices. In this paper we explore possible applications of electric field sensing, a purely passive capacitive measurement technique, which can be implemented with an extremely low power consumption. To cover a wide range of applications, we examine five possible use cases in more detail. The results show that the application is feasible both in interior spaces and outdoors. Moreover, due to the low energy consumption, mobile usage is also possible.
An Observer-Metamerism Sensitivity Index for Electronic Displays
Journal of the Society for Information Display
The effect of observer metamerism induced by electronic displays depends to a large extent on their primary spectra (red, green, and blue in the most common case). In particular, for narrow-band primary spectra whose peak wavelength lies in the range of high variability of the observer's colormatching function, some observers can experience very large differences between actual surface colors (e.g. in a light booth) and displayed colors if the monitor is optimized for the International Commission on Illumination (CIE) 1931 standard observer. However, because narrow-band light-emitting diodes lead to larger color gamuts, more and more monitors with very narrow band primaries are coming onto the market without manufacturers taking into account the associated problem of observer variations. Being able to measure these variations accurately and efficiently is therefore an important objective. In this paper, we propose a new approach to predict the extent of observer metamerism for a particular multiprimary display. Unlike existing dedicated models, ours does not depend on a reference illuminant and a set of reflectance spectra and is computationally more efficient.
Automatic Sentinel Lymph Node Localization in Head and Neck Cancer Using a Coupled Shape Model Algorithm
Computer Assisted and Robotic Endoscopy and Clinical Image-Based Procedures
The localization and analysis of the sentinel lymph node for patients diagnosed with cancer, has significant influence on the prognosis, outcome and treatment of the disease. We present a fully automatic approach to localize the sentinel lymph node and additional active nodes and determine their lymph node level on SPECT-CT data. This is a crucial prerequisite for the planning of radiation therapy or a surgical neck dissection. Our approach was evaluated on 17 lymph nodes. The detection rate of the lymph nodes was 94%; and 88% of the lymph nodes were correctly assigned to their corresponding lymph node level. The proposed algorithm targets a very important topic in clinical practice. The first results are already very promising. The next step has to be the evaluation on a larger data set.
Beyond Group: Multiple Person Tracking via Minimal Topology-Energy-Variation
IEEE Transactions on Image Processing
Tracking multiple persons is a challenging task when persons move in groups and occlude each other. Existing group-based methods have extensively investigated how to make group division more accurately in a tracking-by-detection framework; however, few of them quantify the group dynamics from the perspective of targets' spatial topology or consider the group in a dynamic view. Inspired by the sociological properties of pedestrians, we propose a novel socio-topology model with a topology-energy function to factor the group dynamics of moving persons and groups. In this model, minimizing the topologyenergy- variance in a two-level energy form is expected to produce smooth topology transitions, stable group tracking, and accurate target association. To search for the strong minimum in energy variation, we design the discrete group-tracklet jump moves embedded in the gradient descent method, which ensures that the moves reduce the energy variation of group and trajectory alternately in the varying topology dimension. Experimental results on both RGB and RGB-D data sets show the superiority of our proposed model for multiple person tracking in crowd scenes.
CapSoles: Who Is Walking on What Kind of Floor?
International Conference on Human-Computer Interaction with Mobile Devices and Services (MobileHCI) <19, 2017, Vienna, Austria>
Foot interfaces, such as pressure-sensitive insoles, still yield unused potential such as for implicit interaction. In this paper, we introduce CapSoles, enabling smart insoles to implicitly identify who is walking on what kind of floor. Our insole prototype relies on capacitive sensing and is able to sense plantar pressure distribution underneath the foot, plus a capacitive ground coupling effect. By using machine-learning algorithms, we evaluated the identification of 13 users, while walking, with a confidence of ~95% after a recognition delay of ~1s. Once the user's gait is known, again we can discover irregularities in gait plus a varying ground coupling. While both effects in combination are usually unique for several ground surfaces, we demonstrate to distinguish six kinds of floors, which are sand, lawn, paving stone, carpet, linoleum, and tartan with an average accuracy of ~82%. Moreover, we demonstrate the unique effects of wet and electrostatically charged surfaces.
Contextualization and Recommendation of Annotations to Enhance Information Exchange in Assembly Assistance
J.UCS Journal of Universal Computer Science
Increasingly flexible production processes require intelligent assistance systems containing information and knowledge to maintain high quality and efficiency. To ensure a reliable supply of information, it is of great importance to find easy and fast ways to record and store "new" information, as well as to provide a sensible mechanism to supply the information when needed. In this paper an approach is presented that uses annotations in combination with a formalized knowledge base that represents the work domain. This pre-condition enables a context-based annotation recommendation. A framework is proposed to integrate different factors to measure the relevance of an annotation according to a given situation. The approach is illustrated using the example of an assembly assistance system. To evaluate the users' attitude regarding annotations as instruction support and to test the system's capabilities when handling a great number of annotations some studies were performed and analyzed.
Data Completion in Building Information Management: Electrical Lines from Range Scans and Photographs
Visualization in Engineering
Background: The concept of building information management (BIM) is based on its holistic nature. This idea pays off, if all relevant information is fused into one consistent data set. As a consequence, the completeness of data is vital and the research question on how to complete data automatically remains open. Methods: In this article we present a data completion technique based on knowledge management. We encode expert and domain knowledge in a generative system that represents norms and standards in a machine-readable manner. The implementation of this approach be used to automatically determine a hypothesis on the location of electrical lines within indoor range scans. Results: The generative paradigm can encode domain expert knowledge in a machine-readable way. In this article we demonstrate its usage to represent norms and standards. Conclusions: The benefit of our method is the further completion of digital building information models -- a necessary step to take full advantage of building information modeling.
Developing Knowledge-Based Citizen Participation Platform to Support Smart City Decision Making: The Smarticipate Case Study
Citizen participation for social innovation and co-creating urban regeneration proposals can be greatly facilitated by innovative IT systems. Such systems can use Open Government Data, visualise urban proposals in 3D models and provide automated feedback on the feasibility of the proposals. Using such a system as a communication platform between citizens and city administrations provides an integrated top-down and bottom-up urban planning and decision-making approach to smart cities. However, generating automated feedback on citizens' proposals requires modelling domain-specific knowledge i.e., vocabulary and rules, which can be applied on spatial and temporal 3D models. This paper presents the European Commission funded H2020 smarticipate project that aims to achieve the above challenge by applying it on three smart cities: Hamburg, Rome and RBKC-London. Whilst the proposed system architecture indicates various innovative features, a proof of concept of the automated feedback feature for the Hamburg use case 'planting trees' is demonstrated. Early results and lessons learned show that it is feasible to provide automated feedback on citizen-initiated proposals on specific topics. However, it is not straightforward to generalise this feature to cover more complex concepts and conditions which require specifying comprehensive domain languages, rules and appropriate tools to process them. This paper also highlights the strengths of the smarticipate platform, discusses challenges to realise its different features and suggests potential solutions.
EarFieldSensing: A Novel In-Ear Electric Field Sensing to Enrich Wearable Gesture Input through Facial Expressions
CHI '17. Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems
Conference on Human Factors in Computing Systems (CHI) <35, 2017, Denver, CO, USA>
EarFieldSensing (EarFS) is a novel input method for mobile and wearable computing using facial expressions. Facial muscle movements induce both electric field changes and physical deformations, which are detectable with electrodes placed inside the ear canal. The chosen ear-plug form factor is rather unobtrusive and allows for facial gesture recognition while utilizing the close proximity to the face. We collected 25 facial-related gestures and used them to compare the performance levels of several electric sensing technologies (EMG, CS, EFS, EarFS) with varying electrode setups. Our developed wearable fine-tuned electric field sensing employs differential amplification to effectively cancel out environmental noise while still being sensitive towards small facial-movement-related electric field changes and artifacts from ear canal deformations. By comparing a mobile with a stationary scenario, we found that EarFS continues to perform better in a mobile scenario. Quantitative results show EarFS to be capable of detecting a set of 5 facial gestures with a precision of 90% while sitting and 85.2% while walking. We provide detailed instructions to enable replication of our low-cost sensing device. Applying it to different positions of our body will also allow to sense a variety of other gestures and activities.
Efficient Compression for Server-Side G-Buffer Streaming in Web Applications
Proceedings Web3D 2017
International Conference on 3D Web Technology (WEB3D) <22, 2017, Brisbane, Australia>
Remote rendering methods enable devices with low computing power like smart phones or tablets to visualize massive data. By transmitting G-Buffers, Depth-Image-Based Rendering (DIBR) methods can be used to compensate the artefacts caused by the latency. However, the drawback is that a G-Buffer has at least twice as much data as an image. We present a method for compressing G-Buffers which provides an efficient decoding suitable for web applications. Depending on the computing power of the device, software methods, which run on the CPU, may not be fast enough for an interactive experience. Therefore, we developed a decoding which runs entirely on the GPU. As we use only standard WebGL for our implementation, our compression is suitable for every modern browser.
Efficient, Accurate, and Rotation-Invariant Iris Code
IEEE Signal Processing Letters
The large scale of the recently demanded biometric systems has put a pressure on creating a more efficient, accurate, and private biometric solutions. Iris biometrics is one of the most distinctive and widely used biometric characteristics. High-performing iris representations suffer from the curse of rotation inconsistency. This is usually solved by assuming a range of rotational errors and performing a number of comparisons over this range, which results in a high computational effort and limits indexing and template protection. This work presents a generic and parameter-free transformation of binary iris representation into a rotation-invariant space. The goal is to perform accurate and efficient comparison and enable further indexing and template protection deployment. The proposed approach was tested on a database of 10 000 subjects of the ISYN1 iris database generated by CASIA. Besides providing a compact and rotational-invariant representation, the proposed approach reduced the equal error rate by more than 55% and the computational time by a factor of up to 44 compared to the original representation.
Exercise Monitoring On Consumer Smart Phones Using Ultrasonic Sensing
International Workshop on Sensor-based Activity Recognition (iWOAR) <4, 2017, Rostock, Germany>
Quantified self has been a trend over the last several years. An increasing number of people use devices, such as smartwatches or smartphones to log activities of daily life, including step count or vital information. However, most of these devices have to be worn by the user during the activities, as they rely on integrated motion sensors. Our goal is to create a technology that enables similar precision with remote sensing, based on common sensors installed in every smartphone, in order to enable ubiquitous application. We have created a system that uses the Doppler effect in ultrasound frequencies to detect motion around the smartphone. We propose a novel use case to track exercises, based on several feature extraction methods and machine learning classification. We conducted a study with 14 users, achieving an accuracy between 73% and 92% for the different exercises.
From Reassembly to Object Completion: A Complete Systems Pipeline
ACM Journal on Computing and Cultural Heritage
The problem of the restoration of broken artifacts, where large parts could be missing, is of high importance in archaeology. The typical manual restoration can become a tedious and error-prone process, which also does not scale well. In recent years, many methods have been proposed for assisting the process, most of which target specialized object types or operate under very strict constraints. We propose a digital shape restoration pipeline consisting of proven, robust methods for automatic fragment reassembly and shape completion of generic three-dimensional objects of arbitrary type. In this pipeline, first we introduce a novel unified approach for handling the reassembly of objects from heavily damaged fragments by exploiting both fracture surfaces and salient features on the intact sides of fragments, when available. Second, we propose an object completion procedure based on generalized symmetries and a complementary part extraction process that is suitable for driving the fabrication of missing geometry. We demonstrate the effectiveness of our approach using real-world fractured objects and software implemented as part of the European Union-funded PRESIOUS project, which is also available for download from the project site.
General Borda Count for Multi-biometric Retrieval
2017 International Joint Conference on Biometrics
IEEE International Joint Conference on Biometrics (IJCB) <2017, Denver, CO, USA>
Indexing of multi-biometric data is required to facilitate fast search in large-scale biometric systems. Previous works addressing this issue were challenged by including biometric sources of different nature, utilizing the knowledge about the biometric sources, and optimizing and tuning the retrieval performance. This work presents a generalized multi-biometric retrieval approach that adapts the Borda count algorithm within an optimizable structure. The approach was tested on a database of 10k reference and probe instances of the left and the right irises. The experiments and comparisons to five baseline solutions proved to achieve advances in terms of general indexing performance, tunability to certain operating points, and response to missing data. A clear advantage of the proposed solution was noticed when faced by candidate lists of low quality.
Indexing of Single and Multi-instance Iris Data Based on LSH-Forest and Rotation Invariant Representation
Computer Analysis of Images and Patterns
International Conference on Computer Analysis of Images and Patterns (CAIP) <17, 2017, Ystad, Sweden>
Indexing of iris data is required to facilitate fast search in large-scale biometric systems. Previous works addressing this issue were challenged by the tradeoffs between accuracy, computational efficacy, storage costs, and maintainability. This work presents an iris indexing approach based on rotation invariant iris representation and LSH-Forest to produce an accurate and easily maintainable indexing structure. The complexity of insertion or deletion in the proposed method is limited to the same logarithmic complexity of a query and the required storage grows linearly with the database size. The proposed approach was extended into a multi-instance iris indexing scheme resulting in a clear performance improvement. Single iris indexing scored a hit rate of 99.7% at a 0.1% penetration rate while multi-instance indexing scored a 99.98% hit rate at the same penetration rate. The evaluation of the proposed approach was conducted on a large database of 50k references and 50k probes of the left and the right irises. The advantage of the proposed solution was put into prospective by comparing the achieved performance to the reported results in previous works.
Indoor Localization Based on Passive Electric Field Sensing
European Conference on Ambient Intelligence (AmI) <13, 2017, Malaga, Spain>
The ability to perform accurate indoor positioning opens a wide range of opportunities, including smart home applications and location-based services. Smart floors are a well-established technology to enable marker-free indoor localization within an instrumented environment. Typically, they are based on pressure sensors or varieties of capacitive sensing. These systems, however, are often hard to deploy as mechanical or electrical features are required below the surface. They might also have a limited range or not be compatible with different floor materials. In this paper, we present a novel indoor positioning system using an uncommon form of passive electric field sensing, which detects the change in body electric potential during movement. It is easy to install by deploying a grid of passive wires underneath any non-conductive floor surface. The proposed architecture achieves a high position accuracy and an excellent spatial resolution. In our evaluation, we measure a mean positioning error of only 12.7 cm. The proposed system also combines the advantages of very low power consumption, easy installation, easy maintenance, and the preservation of privacy.
Interactive Physics-Based Deformation for Virtual Worlds
2017 International Conference on Cyberworlds
International Conference on Cyberworlds (CW) <2017, Chester, UK>
When creating immersive interactive virtual worlds, it is important to not only provide plausible visuals, but also to allow the user to interact with the virtual scene in a natural way. While rigid-body physics simulations are widely used to provide basic interaction, realistic soft-body deformations of virtual objects are challenging and therefore typically not offered. We present a client-server architecture for interactive physicsbased deformation, which makes it possible to add physically accurate response to interaction to any virtual environment. The architecture is highly flexible, can be used from any web enabled client, and facilitates synchronization of computed deformations across multiple users and devices.
Interactive Regression Lens for Exploring Scatter Plots
Computer Graphics Forum
Eurographics / IEEE VGTC Conference on Visualization (EuroVis) <19, 2017, Barcelona, Spain>
Data analysis often involves finding models that can explain patterns in data, and reduce possibly large data sets to more compact model-based representations. In Statistics, many methods are available to compute model information. Among others, regression models are widely used to explain data. However, regression analysis typically searches for the best model based on the global distribution of data. On the other hand, a data set may be partitioned into subsets, each requiring individual models. While automatic data subsetting methods exist, these often require parameters or domain knowledge to work with. We propose a system for visual-interactive regression analysis for scatter plot data, supporting both global and local regression modeling. We introduce a novel regression lens concept, allowing a user to interactively select a portion of data, on which regression analysis is run in interactive time. The lens gives encompassing visual feedback on the quality of candidate models as it is interactively navigated across the input data. While our regression lens can be used for fully interactive modeling, we also provide user guidance suggesting appropriate models and data subsets, by means of regression quality scores. We show, by means of use cases, that our regression lens is an effective tool for user-driven regression modeling and supports model understanding.
MATOG: Array Layout Auto-Tuning for CUDA
ACM Transactions on Architecture and Code Optimization
Optimal code performance is (besides correctness and accuracy) the most important objective in compute intensive applications. In many of these applications, Graphic Processing Units (GPUs) are used because of their high amount of compute power. However, caused by their massively parallel architecture, the code has to be specifically adjusted to the underlying hardware to achieve optimal performance and therefore has to be reoptimized for each new generation. In reality, this is usually not the case as productive code is normally at least several years old and nobody has the time to continuously adjust existing code to new hardware. In recent years more and more approaches have emerged that automatically tune the performance of applications toward the underlying hardware. In this article, we present the MATOG auto-tuner and its concepts. It abstracts the array memory access in CUDA applications and automatically optimizes the code according to the used GPUs. MATOG only requires few profiling runs to analyze even complex applications, while achieving significant speedups over non-optimized code, independent of the used GPU generation and without the need to manually tune the code.
MirrorFlow: Exploiting Symmetries in Joint Optical Flow and Occlusion Estimation
2017 IEEE International Conference on Computer Vision
IEEE International Conference on Computer Vision (ICCV) <16, 2017, Venice, Italy>
Optical flow estimation is one of the most studied problems in computer vision, yet recent benchmark datasets continue to reveal problem areas of today's approaches. Occlusions have remained one of the key challenges. In this paper, we propose a symmetric optical flow method to address the well-known chicken-and-egg relation between optical flow and occlusions. In contrast to many state-of-the-art methods that consider occlusions as outliers, possibly filtered out during post-processing, we highlight the importance of joint occlusion reasoning in the optimization and show how to utilize occlusion as an important cue for estimating optical flow. The key feature of our model is to fully exploit the symmetry properties that characterize optical flow and occlusions in the two consecutive images. Specifically through utilizing forward-backward consistency and occlusion-disocclusion symmetry in the energy, our model jointly estimates optical flow in both forward and backward direction, as well as consistent occlusion maps in both views. We demonstrate significant performance benefits on standard benchmarks, especially from the occlusion-disocclusion symmetry. On the challenging KITTI dataset we report the most accurate two-frame results to date.
Nonlinear Statistical Shape Modeling for Ankle Bone Segmentation Using a Novel Kernelized Robust PCA
Medical Image Computing and Computer Assisted Intervention - MICCAI 2017: Part I
International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI) <20, 2017, Québec City, QC, Canada>
Statistical shape models (SSMs) are widely employed in medical image segmentation. However, an inferior SSM will degenerate the quality of segmentations. It is challenging to derive an efficient model because: (1) often the training datasets are corrupted by noise and/or artifacts; (2) conventional SSM is not capable to capture nonlinear variabilities of a population of shape. Addressing these challenges, this work aims to create SSMs that are not only robust to abnormal training data but also satisfied with nonlinear distribution. As Robust PCA is an efficient tool to seek a clean low-rank linear subspace, a novel kernelized Robust PCA (KRPCA) is proposed to cope with nonlinear distribution for statistical shape modeling. In evaluation, the built nonlinear model is used in ankle bone segmentation where 9 bones are separately distributed. Evaluation results show that the model built with KRPCA has a significantly higher quality than other state-of-the-art methods.
ProbFlow: Joint Optical Flow and Uncertainty Estimation
2017 IEEE International Conference on Computer Vision
IEEE International Conference on Computer Vision (ICCV) <16, 2017, Venice, Italy>
Optical flow estimation remains challenging due to untextured areas, motion boundaries, occlusions, and more. Thus, the estimated flow is not equally reliable across the image. To that end, post-hoc confidence measures have been introduced to assess the per-pixel reliability of the flow. We overcome the artificial separation of optical flow and confidence estimation by introducing a method that jointly predicts optical flow and its underlying uncertainty. Starting from common energy-based formulations, we rely on the corresponding posterior distribution of the flow given the images. We derive a variational inference scheme based on mean field, which incorporates best practices from energy minimization. An uncertainty measure is obtained along the flow at every pixel as the (marginal) entropy of the variational distribution. We demonstrate the flexibility of our probabilistic approach by applying it to two different energies and on two benchmarks. We not only obtain flow results that are competitive with the underlying energy minimization approach, but also a reliable uncertainty measure that significantly outperforms existing post-hoc approaches.
Procedural Modeling of Architecture with Round Geometry
Computers & Graphics
International Conference on Cyberworlds (CW) <2016, Chongqing, China>
Creation of procedural 3D building models can significantly reduce the costs of modeling, since it allows for generating a variety of similar shapes from one procedural description. The common field of appli- cation for procedural modeling is modeling of straight building facades, which are very well suited for shape grammars-a special kind of procedural modeling system. In order to generate round building geometry, we present a way to set up different coordinate systems in shape grammars. Besides Cartesian, these are primarily cylindrical and spherical coordinate systems for generation of structures such as towers or domes, that can procedurally adapt to different dimensions and parameters. The users can apply common splitting idioms from shape grammars in their familiar way for creating round instead of straight geometry. The second enhancement we propose is to provide a way for users to give high level inputs that are used to automatically arrange and adapt parts of the models.
Single Image Marine Snow Removal based on a Supervised Median Filtering Scheme
VISAPP 2017. Proceedings
International Conference on Computer Vision Theory and Applications (VISAPP) <12, 2017, Porto, Portugal>
Underwater image processing has attracted a lot of attention due to the special difficulties at capturing clean and high quality images in this medium. Blur, haze, low contrast and color cast are the main degradations. In an underwater image noise is mostly considered as an additive noise (e.g. sensor noise), although the visibility of underwater scenes is distorted by another source, termed marine snow. This signal disturbs image processing methods such as enhancement and segmentation. Therefore removing marine snow can improve image visibility while helping advanced image processing approaches such as background subtraction to yield better results. In this article, we propose a simple but effective filter to eliminate these particles from single underwater images. It consists of different steps which adapt the filter to fit the characteristics of marine snow the best. Our experimental results show the success of our algorithm at outperforming the existing approaches by effectively removing this phenomenon and preserving the edges as much as possible.
State-of-the-Art Overview on 3D Model Representations and Transformations in the Context of Computer-Aided Design
International Journal on Advances in Software
Within a virtual world, either in virtual reality or in a simulation environment, the digital counterparts of real objects are described by mathematical and computational models. Depending on the purpose, the field of application, and the used toolchain a wide variety of model representations is established. As a consequence, conversion methods and transformation algorithms are becoming increasingly important. This article gives a state of the art overview on model representations and on the most important transformation techniques.
Ternary Sparse Matrix Representation for Volumetric Mesh Subdivision and Processing on GPUs
Computer Graphics Forum
Eurographics Symposium on Geometry Processing (SGP) <15, 2017, London, UK>
In this paper, we present a novel volumetric mesh representation suited for parallel computing on modern GPU architectures. The data structure is based on a compact, ternary sparse matrix storage of boundary operators. Boundary operators correspond to the first-order top-down relations of k-faces to their (k-1)-face facets. The compact, ternary matrix storage format is based on compressed sparse row matrices with signed indices and allows for efficient parallel computation of indirect and bottomup relations. This representation is then used in the implementation of several parallel volumetric mesh algorithms including Laplacian smoothing and volumetric Catmull-Clark subdivision. We compare these algorithms with their counterparts based on OpenVolumeMesh and achieve speedups from 3× to 531×, for sufficiently large meshes, while reducing memory consumption by up to 36%.
Copyright: This is the accepted version of the following article: Mueller‐Roemer, J. S., C. Altenhofen, and A. Stork. "Ternary Sparse Matrix Representation for Volumetric Mesh Subdivision and Processing on GPUs." Computer Graphics Forum 36, no. 5 (2017): 59-69, which has been published in final form at http://onlinelibrary.wiley.com. This article may be used for non-commercial purposes in accordance with the Wiley Self-Archiving Policy [http://olabout.wiley.com/WileyCDA/Section/id-820227.html].
Typology of Uncertainty in Static Geolocated Graphs for Visualization
IEEE Computer Graphics and Applications
Static geolocated graphs have nodes connected by edges that can have geographic location and associated attributes. This article proposes a typology of uncertainty in static geolocated graphs, which can affect the existence, location, attributes, or grouping of nodes and edges. The authors also summarize available techniques for visualizing such uncertainty.
Unifying Algebraic Solvers for Scaled Euclidean Registration from Point, Line and Plane Constraints
Computer Vision - ACCV 2016. Part V
Asian Conference on Computer Vision (ACCV) <13, 2016, Taipei, Taiwan>
We investigate recent state-of-the-art algorithms for absolute pose problems (PnP and GPnP) and analyse their applicability to a more general type, namely the scaled Euclidean registration from pointto- point, point-to-line and point-to plane correspondences. Similar to previous formulations we first compress the original set of equations to a least squares error function that only depends on the non-linear rotation parameters and a small symmetric coefficient matrix of fixed size. Then, in a second step the rotation is solved with algorithms which are derived using methods from algebraic geometry such as the Gröbner basis method. In previous approaches the first compression step was usually tailored to a specific correspondence types and problem instances. Here, we propose a unified formulation based on a representation with orthogonal complements which allows to combine different types of constraints elegantly in one single framework. We show that with our unified formulation existing polynomial solvers can be interchangeably applied to problem instances other than those they were originally proposed for. It becomes possible to compare them on various registrations problems with respect to accuracy, numerical stability, and computational speed. Our compression procedure not only preserves linear complexity, it is even faster than previous formulations. For the second step we also derive an own algebraic equation solver, which can additionally handle the registration from 3D point-to-point correspondences, where other solvers surprisingly fail.
Unsupervised Myocardial Segmentation for Cardiac BOLD
IEEE Transactions on Medical Imaging
A fully automated 2-D+time myocardial segmentation framework is proposed for cardiac magnetic resonance (CMR) blood-oxygen-level-dependent (BOLD) data sets. Ischemia detection with CINE BOLD CMR relies on spatio-temporal patterns in myocardial intensity, but these patterns also trouble supervised segmentation methods, the de facto standard for myocardial segmentation in cine MRI. Segmentation errors severely undermine the accurate extraction of these patterns. In this paper, we build a joint motion and appearance method that relies on dictionary learning to find a suitable subspace.Our method is based on variational pre-processing and spatial regularization using Markov random fields, to further improve performance. The superiority of the proposed segmentation technique is demonstrated on a data set containing cardiac phase resolved BOLD MR and standard CINE MR image sequences acquired in baseline and is chemic condition across ten canine subjects. Our unsupervised approach outperforms even supervised state-of-the-art segmentation techniques by at least 10% when using Dice to measure accuracy on BOLD data and performs at par for standard CINE MR. Furthermore, a novel segmental analysis method attuned for BOLD time series is utilized to demonstrate the effectiveness of the proposed method in preserving key BOLD patterns.
Virtual Commissioning of Camera-based Quality Assurance Systems for Mixed Model Assembly Lines
International Conference on Flexible Automation and Intelligent Manufacturing (FAIM) <27, 2017, Modena, Italy>
We propose a new computer vision based technique that does not use data collected from real images to teach a decision-making algorithm. It uses CAD-Models which is already available from the product planning phase to check all product variants on the assembly line for misplaced or wrong components. This solution can hence be used already in the ramp-up phase of new models in which problems are more frequent. Our method can also easily adapt to changes of production, where existing methods need to redo the whole teach-in process, we just need to set the new nominal position of the part in the reference coordinate system, which can even be done semi automatically upfront the changes are initiated in the assembly line.
Virtual Rephotography: Novel View Prediction Error for 3D Reconstruction
ACM Transactions on Graphics
The ultimate goal of many image-based modeling systems is to render photorealistic novel views of a scene without visible artifacts. Existing evaluation metrics and benchmarks focus mainly on the geometric accuracy of the reconstructed model, which is, however, a poor predictor of visual accuracy. Furthermore, using only geometric accuracy by itself does not allow evaluating systems that either lack a geometric scene representation or utilize coarse proxy geometry. Examples include a light field and most image-based rendering systems. We propose a unified evaluation approach based on novel view prediction error that is able to analyze the visual quality of any method that can render novel views from input images. One key advantage of this approach is that it does not require ground truth geometry. This dramatically simplifies the creation of test datasets and benchmarks. It also allows us to evaluate the quality of an unknown scene during the acquisition and reconstruction process, which is useful for acquisition planning. We evaluate our approach on a range of methods, including standard geometry-plus-texture pipelines as well as image-based rendering techniques, compare it to existing geometry-based benchmarks, demonstrate its utility for a range of use cases, and present a new virtual rephotography-based benchmark for image-based modeling and rendering systems.
Visibility-Consistent Thin Surface Reconstruction Using Multi-Scale Kernels
ACM Transactions on Graphics
Conference on Computer and Exhibition on Computer Graphics and Interactive Techniques in Asia (SIGGRAPH ASIA) <10, 2017, Bangkok, Thailand>
One of the key properties of many surface reconstruction techniques is that they represent the volume in front of and behind the surface, e.g., using a variant of signed distance functions. This creates significant problems when reconstructing thin areas of an object since the backside interferes with the reconstruction of the front. We present a two-step technique that avoids this interference and thus imposes no constraints on object thickness. Our method first extracts an approximate surface crust and then iteratively refines the crust to yield the final surface mesh. To extract the crust, we use a novel observation-dependent kernel density estimation to robustly estimate the approximate surface location from the samples. Free space is similarly estimated from the samples' visibility information. In the following refinement, we determine the remaining error using a surface-based kernel interpolation that limits the samples' influence to nearby surface regions with similar orientation and iteratively move the surface towards its true location. We demonstrate our results on synthetic as well as real datasets reconstructed using multi-view stereo techniques or consumer depth sensors.
Visual Interactive Creation and Validation of Text Clustering Workflows to Explore Document Collections
Visualization and Data Analysis 2017
Visualization and Data Analysis (VDA) <2017, Burlingame, CA, USA>
The exploration of text document collections is a complex and cumbersome task. Clustering techniques can help to group documents based on their content for the generation of overviews. However, the underlying clustering workflows comprising preprocessing, feature selection, clustering algorithm selection and parameterization offer several degrees of freedom. Since no "best" clustering workflow exists, users have to evaluate clustering results based on the data and analysis tasks at hand. In our approach, we present an interactive system for the creation and validation of text clustering workflows with the goal to explore document collections. The system allows users to control every step of the text clustering workflow. First, users are supported in the feature selection process via feature selection metrics-based feature ranking and linguistic filtering (e.g., part-of-speech filtering). Second, users can choose between different clustering methods and their parameterizations. Third, the clustering results can be explored based on the cluster content (documents and relevant feature terms), and cluster quality measures. Fourth, the results of different clusterings can be compared, and frequent document subsets in clusters can be identified. We validate the usefulness of the system with a usage scenario describing how users can explore document collections in a visual and interactive way.
Visual-Interactive Similarity Search for Complex Objects by Example of Soccer Player Analysis
IVAPP 2017. Proceedings
International Conference on Information Visualization Theory and Applications (IVAPP) <8, 2017, Porto, Portugal>
The definition of similarity is a key prerequisite when analyzing complex data types in data mining, information retrieval, or machine learning. However, the meaningful definition is often hampered by the complexity of data objects and particularly by different notions of subjective similarity latent in targeted user groups. Taking the example of soccer players, we present a visual-interactive system that learns users' mental models of similarity. In a visual-interactive interface, users are able to label pairs of soccer players with respect to their subjective notion of similarity. Our proposed similarity model automatically learns the respective concept of similarity using an active learning strategy. A visual-interactive retrieval technique is provided to validate the model and to execute downstream retrieval tasks for soccer player analysis. The applicability of the approach is demonstrated in different evaluation strategies, including usage scenarions and cross-validation tests.
Visualization of Delay Uncertainty and its Impact on Train Trip Planning: A Design Study
Computer Graphics Forum
Eurographics / IEEE VGTC Conference on Visualization (EuroVis) <19, 2017, Barcelona, Spain>
Uncertainty about possible train delays has an impact on train trips, as the exact arrival time is unknown during trip planning. Delays can lead to missing a connecting train at the transfer station, or to coming too late to an appointment at the destination. Facing this uncertainty, the traveler may wish to use an earlier train or a different connection arriving well before the appointment. Currently, train trip planning is based on scheduled times of connections between two stations. Information about approximate delays is only available shortly before train departure. Although several visualization approaches can show temporal uncertainty, we are not aware of any visual design specifically supporting trip planning, which can show delay uncertainty and its impact on the connections. We propose and evaluate a visual design which extends train trip planning with delay uncertainty. It shows the scheduled train connections together with their expected train delays as well as their impacts on both the arrival time, and the potential of missing a transfer. The visualization also includes information about alternative connections in case of these critical transfers. In this way the user is able to judge which train connection is suitable for a trip. We conducted a user study with 76 participants to evaluate our design. We compared it to two alternative presentations that are prominent in Germany. The study showed that our design performs comparably well for tasks concerning train schedules. The additional uncertainty display as well as the visualization of alternative connections was appreciated and well understood. The participants were able to estimate when they would likely arrive at their destination despite possible train delays while they were unable to estimate this with existing presentations. The users would prefer to use the new design for their trip planning.
Volumetric Subdivision for Consistent Implicit Mesh Generation
Computers & Graphics
In this paper, we present a novel approach for a tighter integration of 3D modeling and physically- based simulation. Instead of modeling 3D objects as surface models, we use a volumetric subdivision representation. Volumetric modeling operations allow designing 3D objects in similar ways as with surface-based modeling tools, while automatic checks and modifications of inner control points ensure consistency during the design process. Encoding the volumetric information already in the design mesh drastically simplifies and speeds up the mesh generation process for simulation. The transition between design, simulation and back to design is consistent and computationally cheap. Since the subdivision and mesh generation can be expressed as a precomputable matrix-vector multiplication, iteration times can be greatly reduced compared to common modeling and simulation setups. Therefore, this approach is especially well suited for early-stage modeling or optimization use cases, where many geometric changes are made in a short time and their physical effect on the model has to be evaluated frequently. To test our approach, we created, simulated and adapted several 3D models. We measured and evaluated the timings for generating and applying the matrices for different subdivision levels. Additionally, we computed several characteristic factors for mesh quality and mesh consistency. For comparison, we analyzed the tetrahedral meshing functionality offered by CGAL for similar numbers of elements. For changing topology, our implicit meshing approach proves to be up to 70 times faster than creating the tetrahedral mesh only based on the outer surface. Without changing the topology and by precomputing the matrices, we achieve a speed-up of up to 2800, as all the required information is already available.