A5000099153- Advanced Vision and Imaging
- Computer Graphics and Visualization Techniques
- 3D Shape Modeling and Analysis
- Advanced Image Processing Techniques
- Human Pose and Action Recognition
- Advanced Optical Imaging Technologies
- Advanced Image and Video Retrieval Techniques
- Optical measurement and interference techniques
- Data Visualization and Analytics
- Visual Attention and Saliency Detection
- Image Enhancement Techniques
- Video Coding and Compression Technologies
- Human Motion and Animation
- Robotics and Sensor-Based Localization
- Virtual Reality Applications and Impacts
- Image and Video Quality Assessment
- Visual perception and processing mechanisms
- Gaze Tracking and Assistive Technology
- Video Analysis and Summarization
- Augmented Reality Applications
- Medical Image Segmentation Techniques
- 3D Surveying and Cultural Heritage
- Video Surveillance and Tracking Methods
- Image Processing Techniques and Applications
- Sparse and Compressive Sensing Techniques
Technische Universität Braunschweig
2015-2024
University of New Mexico
2013-2021
Max Planck Society
2000-2019
Fraunhofer Institute for Computer Graphics Research
2015
University of British Columbia
2009
Max Planck Institute for Informatics
2003-2007
Hong Kong University of Science and Technology
2007
University of Hong Kong
2007
MPI Research (United States)
2004
Friedrich-Alexander-Universität Erlangen-Nürnberg
1998-2003
In free-viewpoint video, the viewer can interactively choose his viewpoint in 3-D space to observe action of a dynamic real-world scene from arbitrary perspectives. The human body and its motion plays central role most visual media structure be exploited for robust estimation efficient visualization. This paper describes system that uses multi-view synchronized video footage an actor's performance estimate parameters re-render appearance any viewpoint.The silhouettes are extracted frames via...
This paper describes a method to obtain accurate 3D body models and texture of arbitrary people from single, monocular video in which person is moving. Based on parametric model, we present robust processing pipeline infer model shapes including clothed with 4.5mm reconstruction accuracy. At the core our approach transformation dynamic pose into canonical frame reference. Our main contribution transform silhouette cones corresponding human silhouettes visual hull common reference frame....
In free-viewpoint video, the viewer can interactively choose his viewpoint in 3-D space to observe action of a dynamic real-world scene from arbitrary perspectives. The human body and its motion plays central role most visual media structure be exploited for robust estimation efficient visualization. This paper describes system that uses multi-view synchronized video footage an actor's performance estimate parameters re-render appearance any viewpoint.The silhouettes are extracted frames via...
We present Octopus, a learning-based model to infer the personalized 3D shape of people from few frames (1-8) monocular video in which person is moving with reconstruction accuracy 4 5mm, while being orders magnitude faster than previous methods. From semantic segmentation images, our Octopus reconstructs shape, including parameters SMPL plus clothing and hair 10 seconds or less. The achieves fast accurate predictions based on two key design choices. First, by predicting canonical T-pose...
We present a simple yet effective method to infer detailed full human body shape from only single photograph. Our model can full-body including face, hair, and clothing wrinkles at interactive frame-rates. Results feature details even on parts that are occluded in the input image. main idea is turn regression into an aligned image-to-image translation problem. The our partial texture map of visible region obtained off-the-shelf methods. From texture, we estimate normal vector displacement...
We present a novel method for high detail-preserving human avatar creation from monocular video. A parameterized body model is refined and optimized to maximally resemble subjects video showing them all sides. Our avatars feature natural face, hairstyle, clothes with garment wrinkles, high-resolution texture. paper contributes facial landmark shading-based shape refinement, semantic texture prior, stitching strategy, resulting in the most sophisticated-looking obtained single date. Numerous...
The eight articles in this special section are devoted to multi-view imaging and three dimensional television displays.
Abstract This state of the art report covers reconstruction methods for transparent and specular objects or phenomena. While 3D acquisition opaque surfaces with Lambertian reflectance is a well‐studied problem, transparent, refractive, potentially dynamic scenes pose challenging problems systems. reviews categorizes literature in this field. Despite tremendous interest object digitization, digital models far from being solved problem. On other hand, real‐world data high demand applications...
Fluid simulation is one of the most active research areas in computer graphics. However, it remains difficult to obtain measurements real fluid flows for validation simulated data. In this paper, we take a step direction capturing flow data such purposes. Specifically, present first time-resolved Schlieren tomography system full 3D, non-stationary gas on dense volumetric grid. uses 2D ray deflection reconstruct time-varying grid 3D refractive index values, which directly correspond physical...
We present a flexible interactive 3D morpho-kinematical modeling application for astrophysics. Compared to other systems, our reduces the restrictions on physical assumptions, data type, and amount that is required reconstruction of an object's morphology. It one first publicly available tools apply graphics astrophysical modeling. The tool allows astrophysicists provide priori knowledge about object by interactively defining structural elements. By direct comparison model prediction with...
We propose a method that extracts sparse and spatially localized deformation modes from an animated mesh sequence. To this end, we new way to extend the theory of matrix decompositions 3D sequence processing, further contribute with automatic ensure spatial locality decomposition in optimization framework. The extracted dimensions often have intuitive clear interpretable meaning. Our optionally accepts user-constraints guide process discovering underlying latent space. capabilities our...
We present a method to analytically compute the light distribution of triangles directly in frequency space. This allows for fast evaluation, shading, and propagation from 3D mesh objects using angular spectrum methods. The algorithm complexity is only dependent on hologram resolution polygon count model. In contrast other based computer generated holography methods we do not need perform Fourier transform per surface. theory behind approach derived, suitable digital general triangle...
Visual exploration of multivariate data typically requires projection onto lower-dimensional representations. The number possible representations grows rapidly with the dimensions, and manual quickly becomes ineffective or even unfeasible. This paper proposes automatic analysis methods to extract potentially relevant visual structures from a set candidate visualizations. Based on features, visualizations are ranked in accordance specified user task. is provided manageable useful...
Two light-field compression schemes are presented. The codecs compared with regard to efficiency and rendering performance. first proposed coder is based on video-compression techniques that have been modified code the four-dimensional data structure efficiently. second relies entirely disparity-compensated image prediction, establishing a hierarchical among images. coding performance of both evaluated using publicly available light fields synthetic, as well real-world, scenes. Compression...
Abstract We present a novel multi‐view, projective texture mapping technique. While previous multi‐view texturing approaches lead to blurring and ghosting artefacts if 3D geometry and/or camera calibration are imprecise, we propose algorithm that warps (“floats”) projected textures during run‐time preserve crisp, detailed appearance. Our GPU implementation achieves interactive real‐time frame rates. The method is very generally applicable can be used in combination with many image‐based...
An algorithmic framework to compute sparse or minimal-TV solutions of linear systems is proposed. The includes both the Kaczmarz method and linearized Bregman as special cases also several new methods such a solver. has variety applications especially useful for problems in which measurements are slow expensive obtain. We present examples online compressed sensing, TV tomographic reconstruction radio interferometry.
With ever-increasing display resolution for wide field-of-view displays—such as head-mounted displays or 8k projectors—shading has become the major computational cost in rasterization. To reduce effort, we propose an algorithm that only shades visible features of image while cost-effectively interpolating remaining without affecting perceived quality. In contrast to previous approaches do not simulate acuity falloff but also introduce a sampling scheme incorporates multiple aspects human...
We investigate the feasibility of combining off-the-shelf virtual reality headsets and electroencephalography. EEG is a highly sensitive tool subject to strong distortions when exerting physical force like mounting VR headset on top it that twists sensors cables. Our study compares signal quality in against immersive dome environments traditional displays using an oddball paradigm experimental design. Furthermore, we compare combined with commodity without modification modified version...
This paper presents a novel method for using programmable graphics hardware to generate fringe patterns SLM-based holographic displays. The algorithm is designed take the programming constraints imposed by pipeline model into consideration, and scales linearly with number of object points. In contrast previous methods we do not have use Fresnel approximation. technique can also be used on several processors in parallel further optimization. We achieve real-time frame rates objects consisting...
Non-invasively determining the three-dimensional structure of real flames is a challenging task. We present tomographic method for reconstructing volumetric model from multiple images fire. The similar to sparse-view computerized tomography and applicable static camera setups observing dynamic flames. Using an algebraic reconstruction method, we can restrict solution space such that high quality obtained only small number images. An additional advantage fast processing multi-video sequences...
Abstract The 2007 DARPA Urban Challenge afforded the golden opportunity for Technische Universität Braunschweig to demonstrate its abilities develop an autonomously driving vehicle compete with world's best. After several stages of qualification, our team CarOLO qualified early Final Event and was among only 11 teams from initially 89 competitors in final. We had ability work together a large group experts, each contributing his expertise discipline, significant organizational, financial,...
Modern visualization methods are needed to cope with very high-dimensional data. Efficient visual analytical techniques required extract the information content in these The large number of possible projections for each method, which usually grow quadrat-ically or even exponentially dimensions, urges necessity employ automatic reduction techniques, sorting selecting projections, based on their information-bearing content. Different quality measures have been successfully applied several...
Abstract Interactive computation of global illumination is a major challenge in current computer graphics research. Global heavily affects the visual quality generated images. It therefore key attribute for perception photo‐realistic Path tracing able to simulate physical behaviour light using Monte Carlo techniques. However, computational burden this technique prohibits interactive rendering times on standard commodity hardware high‐quality. Trying solve integration with fewer samples...
Visual exploration of multivariate data typically requires projection onto lower dimensional representations. The number possible representations grows rapidly with the dimensions, and manual quickly becomes ineffective or even unfeasible. This paper proposes automatic analysis methods to extract potentially relevant visual structures from a set candidate visualizations. Based on features, visualizations are ranked in accordance specified user task. is provided manageable useful...