Eurographics Partner Events

Permanent URI for this community

https://diglib.eg.org/handle/10.2312/974

Browse

Now showing 1 - 20 of 2608

Frontiers in 3D Photography: Reflectance and Motion
(The Eurographics Association, 2003) Seitz, Steve; Peter Hall and Philip Willis
The last decade has seen great progress in 3D shape capture technology, both in terms of research ad- vances and successful commercial systems. So why don't most people own 3D cameras? Aside from such issues as expense and size, there remain fundamental technical problems that restrict the applicability of the current state of the art. One of these problems is modeling realistic materials. If you look around, you'll notice that most objects are shiny and have specular highlights. However, virtually all shape recovery methods perform poorly for such surfaces. For this reason, it is common to paint an object white before scanning with a laser scanner. This limitation seems unnecessary, given that the human visual system appears to have no such problem{in fact, highlights should give us more information for shape inference. Another major limitation is that the 3D capture process is currently slow and laborious. In particular, very few techniques exist that can capture moving scenes. Obtaining accurate and complete models of moving scenes is challenging, due to the limited measurements that can be obtained at each instant in time. In this talk, I will describe new work from my research group that seeks to address these two problems{ modeling shiny scenes and moving scenes from photographs. We have recently developed shape capture methods that produce laser-scanner-quality results but for scenes with general re ective properties, including shiny surfaces and even anisotropic materials like brushed fur. I will also describe motion- scanning techniques that yield dense shape reconstructions of moving scenes.
Real-Time Capture, Reconstruction and Insertion into Virtual World of Human Actors
(The Eurographics Association, 2003) Hasenfratz, J.M.; Lapierre, M.; Gascuel, J.-D.; Boyer, E.; Peter Hall and Philip Willis
In this paper, we show how to capture an actor with no intrusive trackers and without any special environment like blue set, how to estimate its 3D-geometry and how to insert this geometry into a virtual world in real-time. We use several cameras in conjunction with background subtraction to produce silhouettes of the actor as observed from the different camera viewpoints. These silhouettes allow the 3D-geometry of the actor to be estimated by a voxel based method. This geometry is rendered with a marching cube algorithm and inserted into a virtual world. Shadows of the actor corresponding to virtual lights are then added and interactions with objects of the virtual world are proposed. The main originality of this paper is to propose a complete pipeline that can computes up to 30 frames per second. Since the rapidity of the process depends mainly on its slowest step, we present here all these steps. For each of them, we present and discuss the solution that is used. Some of them are new solutions, as the 3D shape estimation which is achieved using graphics hardware. Results are presented and discussed.
A Framework for Dynamic Connectivity Meshes
(The Eurographics Association, 2003) J.Vorsatz,; Seidel, Hans-Peter; Dirk Reiners
Implementing algorithms that are based on dynamic triangle meshes often requires updating internal datastructures as soon as the connectivity of the mesh changes. The design of a class hierarchy that is able to deal with such changes is particularly challenging if the system reaches a certain complexity. The paper proposes a software design that enables the users to efficiently implement algorithms that can handle these dynamic changes while still maintaining a certain encapsulation of the single components. Our design is based on a callback mechanism. A client can register at some Info-object and gets informed whenever a change of the connectivity occurs. This way the client is able to keep internal data up-to-date. Our framework enables us to write small client classes that cover just a small dedicated aspect of necessary updates related to the changing connectivity. These small components can be combined to more complex modules and can often easily be reused. Moreover, we do not have to store related 'dynamic data' in one central place, e.g. the mesh, which could lead to a significant memory overhead if an application uses some modules just for a short time. We have used and tested this class design extensively for implementing 'Dynamic Connectivity Meshes and Applications 9'. Additionally, as a feasibility study, we have implemented and integrated our concept in the OpenMesh2- framework.
A Novel Form of Pointing Device
(The Eurographics Association, 2003) Cantzler, H.; Hoile, C.; Peter Hall and Philip Willis
This paper presents a novel approach for man machine interaction applying real time computer vision techniques. We use a handheld camera to control the mouse cursor on a computer display. The camera captures an image of the display in its eld of view and this can be used to judge the camera's position and orientation relative to the display. The problem is modelled as a plane-to-plane projection (homography). Once the mapping of the display in the camera view to the real world display is known, the intersection between the central axis of the camera and the surface of the display can be computed. The mouse pointer is then moved to the corresponding display position. This calculation can be iterated continuously to update the mouse cursor position as the camera position and orientation changes. The camera can then be used to control the mouse cursor just like a laser pointer controls a laser dot. A prototype has been developed to demonstrate the approach.
Use and Re-use of Facial Motion CaptureData
(The Eurographics Association, 2003) Lorenzo, M.S.; Edge, J.D.; King, S.A.; Maddock, S.; Peter Hall and Philip Willis
Motion capture (mocap) data is commonly used to recreate complex human motions in computer graphics. Markers are placed on an actor, and the captured movement of these markers allows us to animate computer-generated characters. Technologies have been introduced which allow this technique to be used not only to retrieve rigid body transformations, but also soft body motion such as the facial movement of an actor. The inherent difficulties of working with facial mocap lies in the application of a discrete sampling of surface points to animate a fine discontinuous mesh. Furthermore, in the general case, where the morphology of the actor's face does not coincide with that of the model we wish to animate, some form of retargetting must be applied. In this paper we discuss methods to animate face meshes from mocap data with minimal user intervention using a surface-oriented deformation paradigm.
An Eigenvector Method for Surface Recovery
(The Eurographics Association, 2003) Robles-Kelly, A.; Hancock, E.R.; Peter Hall and Philip Willis
In this paper we explore how spectral methods for graph seriation can be used to develop a new shape-fromshading algorithm. We characterise the field of surface normals using a transition matrix whose elements are computed from the sectional curvature between different image locations. We use a graph seriation method to define a curvature minimising surface integration path for the purposes of height reconstruction. To smooth the reconstructed surface, we fit quadric patches to the height data. The smoothed surface normal directions are updated ensuring compliance with Lambert's law. The processes of height recovery and surface normal adjustment are interleaved and iterated until a stable surface is obtained. We provide results on synthetic and real-world imagery.
Parallel Architecture of an Interactive Scientific Visualisation System for Large Datasets
(The Eurographics Association, 2003) Schneider, Sascha; May, Thorsten; Schmidt, Michael; Dirk Reiners
In this paper we describe a further development state of our system which is able to compute actual scientific and realistic visualization methods in parallel. This paper is related to the basic work we presented in one of our previous articles 18. Our system is capable to integrate easily in modern VR renderers like for example Open Inventor 19, Coin 4 and OpenSG 5. Our approach is designed for processing large datasets which usually are the result of physically based simulation algorithms and programs. Using our techniques it is even more feasible to manage similar visualization problems for other large amounts of data (e.g. medicinal CT-scans or large geometries) in the context of displaying interactively.
Prometheus: Facial Modelling, Tracking and Puppetry
(The Eurographics Association, 2003) Thorne, J.M.; Chatting, D.J.; Peter Hall and Philip Willis
The Prometheus project sought to create a real-time production chain for 3D content. This paper summarises the techniques that were developed for facial modelling, tracking and puppetry. The process used for creating photo-realistic models from images of the face is discussed along with the creation of a real-time markerless facial feature tracker, highlighting the methods for extracting pose, detecting the shape of the mouth and interpreting the occurrence of skin wrinkling. The conversion of tracking data to useable animation is also addressed.
Lambertian Correction for Rough and Specular Surfaces
(The Eurographics Association, 2003) Robles-Kelly, A.; Hancock, E.R.; Peter Hall and Philip Willis
This paper describes a method for performing Lambertian reflectance for rough and specular surfaces. Rather than using an existing reflectance model, we present a method for estimating the reflectance function from image data. The method makes use of the Gauss map between a surface and a unit sphere. Under conditions in which the light source direction and the viewer direction are identical, we show how the reflectance function can be represented by a polar function on the unit sphere. We pose the problem of recovering the reflectance function as that of estimating a tabular representation of the polar function. A simple analysis shows how the tabular representation of the reflectance function can be obtained using the accumulative distribution of image gradients. By modifying the reflectance function and back-projecting, we can render the surface with alternative lighting models. Here, we choose to back-project a Lambertian reflectance model. This allows us to be remove specularities from shiny surfaces and compensate from boundary ''flattening'' for rough surfaces. We illustrate the utility of the method on a variety of real world imagery.
Preserving Realism in Real-Time Rendering of Bidirectional Texture Functions
(The Eurographics Association, 2003) Meseth, Jan; Müller, Gero; Klein, Reinhard; Dirk Reiners
The Bidirectional Texture Function (BTF) is a suitable representation for the appearance of highly detailed surface structures under varying illumination and viewing conditions. Since real-time rendering of the full BTF data is currently not feasible, approximations of the six-dimensional BTF are used such that the amount of data is reduced and current graphics hardware can be exploited. While existing methods work well for materials with low depth variation, realism is lost if the depth variation grows. In this paper we analyze this problem and devise a new real-time rendering method, which provides signi cant improvements with respect to realism for such highly structured materials without sacri cing the general applicability and speed of previous algorithms. We combine our approach with texture synthesis methods to drastically reduce the texture memory requirements and demonstrate the capabilities of our new rendering method with several examples.
A Vision-Based Location System using Fiducials
(The Eurographics Association, 2003) Johnston, D.J.; Clark, A.F.; Peter Hall and Philip Willis
A system for vision-based ego location using 'targets' or 'fiducials' is described. The system is robust and operates on commodity hardware in real time. The accuracy of the system is assessed and found to be good enough to support some augmented reality applications. Two example applications are described: the control of an avatar in a shared virtual environment and a video 'joystick' for manipulating 3D models.
Iterative Multi-Planar Camera Calibration: Improving Stability using Model Selection
(The Eurographics Association, 2003) Vigueras, J.F.; Berger, M.-O.; Simon, G.; Peter Hall and Philip Willis
Tracking, or camera pose determination, is the main technical challenge in numerous applications in computer vision and especially in Augmented Reality. However, pose computation processes commonly exhibit some fluctuations and lack of precision in the estimation of the parameters. This leads to unpleasant visual impressions when augmented scenes are considered. In this paper, we propose an efficient and reliable method for real time camera tracking which avoid unpleasant statistical fluctuations. This method is based on the knowledge of a piecewise planar structure in the scene and makes use of model selection to reduce fluctuations. Videos are attached to this paper which prove the effectiveness of our approach.
A Region Adjacency Tree Approach to the Detection and Design of Fiducials
(The Eurographics Association, 2003) Costanza, E.; Robinson, J.; Peter Hall and Philip Willis
We report a topological approach to fiducial recognition for real-time applications. Independence from geometry makes the system tolerant to severe distortion, and allows encoding of extra information. The method is based on region adjacency trees. After describing the mathematical foundations, we present a set of simulations to evaluate the algorithm and optimise the fiducial design.
Collaborative Vision and Interactive Mosaicing
(The Eurographics Association, 2003) Robinson, J.A.; Peter Hall and Philip Willis
I propose criteria for collaborative vision applications where a camera user/operator and a computer work together to analyse a scene. An example of how these may be fulfilled is provided in IMP - an interactive mosaicing program.IMP generates mosaics in real-time, interacting with the user to cue camera movement and relay performance in-formation.
An Occlusion Culling Toolkit for OpenSG PLUS
(The Eurographics Association, 2003) Staneker, Dirk; Dirk Reiners
Image-space occlusion culling is an useful approach to reduce the rendering load of large polygonal models in scienti c computing, mechanical engineering, or virtual medicine. Like most large model techniques, occlusion culling trades overhead costs with the rendering costs of the possibly occluded geometry. In this paper we present an occlusion culling toolkit for OpenSG. The toolkit includes three different image space techniques utilizing graphics hardware. These techniques are supported by other software techniques to optimize the occlusion culling queries. All techniques are conservative and usable with dynamic scenes, because no precomputing is necessary.
Extending Natural Textures with Multi-Scale Synthesis
(The Eurographics Association, 2003) Stahlhut, O.; Peter Hall and Philip Willis
This paper presents a texture synthesis algorithm that was designed for the tile-less generation of large images of arbitrary size from small sample images. The synthesised texture shows features that are visually similar to the sample over a wide frequency range. The development of the algorithm aimed at achieving high quality results for a large range of natural textures, incorporation of the original samples in the synthesis product, ease of use and good texturing speed even with input sample data two magnitudes larger than used by previous techniques. Like other algorithms we utilise an implicit texture model by copying arbitrary shaped texture patches from the sample to the destination over a multi-scale image pyramid. Our method combines the advantages of different previous techniques with respect to quality. A mixture of exhaustive searching, massive parallel computing and the well-known LBG-algorithm ensures a good balance between texturing quality and speed.
User Interfaces for Mobile Augmented Reality Systems
(The Eurographics Association, 2003) Feiner, Steve; Peter Hall and Philip Willis
What should user interfaces look like when they become an integral part of how we experience the world around us? This talk provides an overview of work that explores user interface design issues for mobile augmented reality systems, which use tracked see-through and hear-through displays to overlay virtual graphics and sound on the real world.
Visualising Video Sequences using Direct Volume Rendering
(The Eurographics Association, 2003) Daniel, G.; Chen, M.; Peter Hall and Philip Willis
It is evident that more and more video data is being generated everyday, for example, by TV broadcast companies and security cameras. However, whilst we are overwhelmed by the huge amount of imagery data, machine vision is generally not yet ready to replace us in most of the everyday visual tasks. In this paper we present a novel approach to the handling of video data. We propose to employ volume visualisation techniques for ''summarising'' video sequences, and to render video volumes into appropriate visual representations. Such visualisations can be used to assist in a decision making process, for instance, to determine if there is any unusual level of activity recorded in a video. In the paper, we present a prototype system, called V3, for capturing, managing, processing and visualising video data sets. We highlight the conceptual similarity between video visualisation and the traditional volume visualisation, and describe the deployment of conventional transfer functions and spatial transfer functions in video visualisation. We examine several statistical image comparison metrics and discuss their effectiveness in extracting meaningful information from video sequences. This work demonstrates the importance and the potential of combining graphics, video and vision technologies.
Models from Image Triplets using Epipolar Gradient Features
(The Eurographics Association, 2003) Vincent, É.; Laganière, R.; Peter Hall and Philip Willis
In an application where sparse matching of feature points is used towards fast scene reconstruction, the choice of the type of features to be matched has an important impact on the quality of the resulting model. In this work, a method is presented for quickly and reliably selecting and matching points from three views of a scene. The selected points are based on epipolar gradients, and consist in stable image features relevant to reconstruction. Then, the selected points are matched using edge transfer, a measure of geometric consistency for point triplets and the edges on which they lie. This matching scheme is invariant to image deformations due to changes in viewpoint. Models drawn from matches obtained by the proposed technique are shown to demonstrate its usefulness.
Towards A 3D Virtual Studio for Human Appearance Capture
(The Eurographics Association, 2003) Starck, J.; Hilton, A.; Peter Hall and Philip Willis
This paper introduces the concept of a ''3D Virtual Studio'' for human appearance capture, akin to the motion capture studio for human motion tracking. Ultimately the 3D Virtual Studio should enable video-realistic reconstruction of a moving person from any viewpoint. A mesh-based stereo technique is presented to reconstruct a moving person from multiple camera views. This technique optimises a surface mesh to match stereo and silhouette data in a constrained coarse-to-fine framework, recovering sub-pixel image correspondence in the presence of inexact camera calibration. We compare this approach for scene reconstruction to conventional shape from silhouette and multiple view stereo. We then demonstrate view-dependent rendering and show improved resolution with the recovered image correspondence. We then demonstrate how this approach can be used to capture the dynamic shape and appearance of a computer graphics model of a person.

Browse

Browsing Eurographics Partner Events by Issue Date

Results Per Page

Sort Options