WSCG '2004: Short Communications

Permanent URI for this collection

http://hdl.handle.net/11025/6177

Browse

Showing 1 - 20 out of 40 results

Segmentation of Hemodynamics from Dynamic- Susceptibility-Contrast Magnetic Resonance Brain Images Using Sequential Independent Component Analysis
(Václav Skala - UNION Agency, 2004) Wu, Yu-Te; Chen, Hui-Yun; Hung, Chih-I; Yi-Hsuan, Kao; Guo, Wan-Yuo; Yeh, Tzu-Chen; Hsieh, Jen-Chuen; Skala, Václav
Dynamic-susceptibility-contrast magnetic resonance imaging, a popular perfusion imaging technique, records signal changes on images caused by the passage of contrast-agent particles in the human brain after a bolus injection of contrast agent. The temporal signal changes on different brain tissues characterize distinct blood supply patterns which are critical for the profound analysis of cerebral hemodynamics. Under the assumption of the spatial independence among these patterns, independent component analysis (ICA) was applied to segment different tissues, i.e., artery, gray matter, white matter, vein and sinus and choroids plexus, so that the spatio- temporal hemodynamics of these tissues were decomposed and analyzed. An arterial input function was modeled using the concentration-time curve of the arterial area for the deconvolution calculation of relative cerebral blood flow. The cerebral hemodynamic parameters, such as relative cerebral blood volume (rCBV), relative cerebral blood flow (rCBF), and relative mean transit time (rMTT), were computed and their averaged ratios between gray matter and white matter were in good agreement with those in the literature.
Non Conventional 3D Human Face Verification System
(Václav Skala - UNION Agency, 2004) Zimmermann, A. C.; Encinas, L. S.; Gonçalves, A. A.; Barreto, J. M.; Skala, Václav
Improving Depth Maps by Nonlinear Diffusion
(Václav Skala - UNION Agency, 2004) Yin, Jianfeng; Cooperstock, Jeremy R.; Skala, Václav
Dense depth maps, typically produced by stereo algorithms, are essential for various computer vision applications. For general configurations in which the cameras are not necessarily parallel or close together, it often proves difficult to obtain reasonable results for complex scenes, in particular in occluded or textureless regions. To improve the depth map in such regions, we propose a post-processing method and illustrate its benefits to applications such as 3D reconstruction or foreground segmentation of a user in a scene.
Blending implicit shapes using smooth unit step functions
(Václav Skala - UNION Agency, 2004) Li, Qingde; Skala, Václav
In constructive solid geometry (CSG), highly smooth and manageable blending operations are always desirable. In this paper, we propose two novel ways to blend implicitly defined geometric objects using smooth unit step func- tions. With the proposed techniques, not only is it direct to build an implicit blending operation with controllable blending range, but also the blending operation can be constructed to whatever level of smoothness. In addition, the mathematical descriptions of the techniques are quite simple and elegant.
Cinematic Camera Control in 3D Computer Games
(Václav Skala - UNION Agency, 2004) Ling-Chieh, Lin; Zen-Chung, Shih; Tsai, Yu-Ting; Skala, Václav
Good camera control and planning techniques in 3D comput er games can give players deeper feelings about atmosphere of games. However, most modern computer games use fixed point-of-view techniques to control the scene camera. We believe applications of some basic cinematic camera planning methods in computer games will enrich player’s gaming experience. In this paper, we discuss the principles of camera control in real-time 3D computer games and concepts of cinematography, encapsulating both of them into camera modules. We also present a system for event-driven camera modules selecti on which is connected to the interactive environment in 3D computer games. Furthermore, some methods to solve the conflict between camera placement and frame coherence are also described. Finally, a practical method to avoid obstruction is used in this system to consider the visibility of important objects.
Hierarchical disparity estimation with programmable 3D hardware
(Václav Skala - UNION Agency, 2004) Zach, Christopher; Karner, Konrad; Bischof, Horst; Skala, Václav
This work presents an implementation of a hierarchical disparity estimation algorithm entirely executed on programmable 3D graphics hardware. In contrast to previous hardware based implementations of computational stereo algorithms, our method calculates disparities either for rectified stereo images or uncalibrated pairs of stereo images without known epipolar geometry. We exploit features of modern graphics hardware to search for correct disparity vectors efficiently. The hierarchical approach increases the speed and the robustness of the algorithm. Additionally, we use bidirectional matching to remove false matches. We observe up to 50 fps for input images with 256×256 pixels resolution on desktop graphics cards and up to 30 fps on mobile 3D hardware.
GPU-based 3D texture advection for the visualization of unsteady flow fields
(Václav Skala - UNION Agency, 2004) Weiskopf, Daniel; Ertl, Thomas; Skala, Václav
We present an interactive visualization approach for the dense representation of unsteady 3D flow fields. The first part of this approach is a GPU-based 3D texture advection scheme that allows a slice of the 3D visual representation to be updated in a single rendering pass. In the second step, the result of the advection process is displayed by texture-based volume rendering. Since both parts are completely supported by the GPU, interactive frame rates are achieved for the visualization of time-dependent flow fields. Moreover, the noise and dye injection scheme of Image Based Flow Visualization (IBFV) is adopted and generalized to take into account a flexible combination of advected and newly injected values. In addition, the advection and rendering methods are extended to transport and display different materials instead of the aggregated colors and opacities. This approach leads to a unified description of noise and dye advection and allows the user to specifically emphasize or blend out regions of the flow.
User-defined texture synthesis
(Václav Skala - UNION Agency, 2004) Taponecco, Francesca; Skala, Václav
Synthesis of textures is a very popular and active area of research; the applications and the areas of interest are various and significant. In the last years, much work has been done in order to optimize the synthesis process, speeding up the methods and minimizing the processing errors. Recent efforts in this area are concentrated in producing flexible algorithms and in introducing tools, which augment the texture with artistic effects. In this work we present a novel approach for flexible texture synthesis: starting from an input sample the process generates an arbitrary resolution output image; the characteristics of this output texture are user-defined, as the user can freely choose a force field, determine color variations and add further features. The synthesis process is fully automatic and does not require additional intervention. The resulting outputs can be interpreted as filtered versions of a texture or as being obtained through transfer functions.
Automatic fitting and control of complex freeform shapes
(Václav Skala - UNION Agency, 2004) Song, Y.; Vergeest, J. S. M.; Wang, C.; Skala, Václav
In many computer graphics and computer-aided design problems, it is very common to find a smooth and well structured surface to fit a set of unstructured 3-dimensional data. Although general approaches of fitting give satisfactory results, the computation time and the complexity often prevent their further developments in more complex cases especially in reusing an existing design. In this paper, for a better control of existing freeform shapes, they are approximated by feature templates, with emphasis on extendable templates. By the advantage of the small number of intrinsic parameters in the feature based deformable templates, fitting procedures are faster and more robust. Three key types of simple freeform templates, the bump, the ridge and the hole, are introduced first. With distance measuring methods, a uniform optimization function is presented to achieve automatic feature recognition and fitting. By introducing the extendable template, the hole and the ridge template are further developed to match complex freeform shapes. Based on the approximated template, further shape manipulations can be conducted effectively using the shape intrinsic parameters. Numerical experiments are conducted in order to verify the proposed algorithms. It is also described how the matching technique can be applied in computer graphics and computer-aided design applications.
Evaluation of NURBS surfaces: an overview based on runtime efficiency
(Václav Skala - UNION Agency, 2004) Sánchez, H.; Moreno, A.; Oyarzun, D.; Gárcía-Alonso, A.; Skala, Václav
Dynamic tessellation of (trimmed) NURBS surfaces is an important topic for interactive rendering of complex models. Independently of the decomposition method used to approximate surfaces by polygons, an evaluation of the surface at many parametric pairs is required to generate meshes vertices. In this paper we give an overview of the most important evaluation algorithms and compare them regarding their runtime efficiency. Three different representations and five algorithms for the evaluation of NURBS surfaces are compared: direct evaluation of NURBS surfaces, evaluation of their Bezier patches and evaluation of them in power basis representation. We add to the comparison two algorithms, based on the approximated computation of normal vectors. The deviation among the exact and the approximated normal vectors is measured. The obtained results are checked with partial comparisons that appear in the bibliography.
Voxelization of solids using simplicial coverings
(Václav Skala - UNION Agency, 2004) Rueda, Antonio J.; Segura, Rafael J.; Feito, Francisco R.; de Miras, Juan Ruiz; Ogáyar, Carlos; Skala, Václav
Rasterization of polygons in 2D is a well known problem, existing several optimal solutions to solve it. The extension of this problem to 3D is more difficult and most existing solutions are designed to obtain a voxelization of the solid. In this paper a new approach to rasterize and voxelize solids in 3D is presented. The described algorithms are very simple, general and robust. The 3D algorithm is valid to be used in the new 3D displays, and it can also be used to voxelize solids delimited by planar faces (with or without holes, manifold or non-manifold). The proposed methods are very suitable for an implementation in graphic hardware rendering system, because it does not use any additional data structure or complex operation.
Oriented bounding boxes using multiresolution contours for fast interference detection of arbitrary geometry objects
(Václav Skala - UNION Agency, 2011) Rivera, Luis; Estrela, Vania; Carvalho, Paulo C. P.; Skala, Václav
Interference detection of arbitrary geometric objects is not a trivial task due to the heavy computational load imposed by implementation issues. The hierarchically structured bounding boxes help us to quickly isolate the contour of segments in interference. In this paper, a new approach is introduced to treat the interference detection problem involving the representation of arbitrary shaped objects. Our proposed method relies upon searching for the best possible way to represent contours by means of hierarchically structured rectangular oriented bounding boxes. This technique handles 2D objects boundaries defined by closed B-spline curves with roughness details. Each oriented box is adapted and fitted to the segments of the contour using second order statistical indicators from some elements of the segments of the object contour in a multiresolution framework. Our method is efficient and robust when it comes to 2D animations in real time. It can deal with smooth curves and polygonal approximations as well results are present to illustrate the performance of the new method.
Efficient implementation of higher order image interpolation
(Václav Skala - UNION Agency, 2010) Perko, Roland; Bischof, Horst; Skala, Václav
This work presents a new method of fast cubic and higher order image interpolation. The evaluation of the piecewise n-th order polynomial kernels is accelerated by transforming the polynomials into the interval [0,1], which has the advantage that some terms of the polynomials disappear, and that several coefficients could be precalculated, which is proven in the paper. The results are exactly the same as using standard n-th order interpolation, but the computational complexity is reduced. Calculating the interpolation weights for the cubic convolution only needs about 60% of the time compared to the classical method optimized by the Horner's rule. This allows a new efficient implementation for image interpolation.
Partial visibility for virtual reality application
(Václav Skala - UNION Agency, 2009) Pearce, D.; Day, A M.; Skala, Václav
This paper presents new approximated visibility algorithms. The aim is to develop output sensitive algorithms for virtual environment walk-through applications. We aim to achieve efficient rendering of complex computer models containing partially occluded areas. Emphasis is paid to the rendering of natural environment models. The algorithms presented in this paper perform visibility calculations in a plane, these algorithms have a more general 3D analogy which has many practical applications within the field of virtual environments. The algorithms presented here can be used with minor extensions to be applied to 2.5D ground based virtual environment walk-throughs.
Extraction of volumetric structures in an illuminance image
(Václav Skala - UNION Agency, 2008) Mokhtari, Marielle; Bergevin, Robert; Skala, Václav
An original method is proposed to extract the most significant volumetric structures in an illuminance im age. The method proceeds in three levels of organization managed by generic grouping principles: (i) from the illuminance im age to amore compact representation of its contents by generic structural in formation extraction leading to a basic contour primitive map; (ii) grouping of the basic primitives in order to form intermediate primitives, the contour junctions; (iii) grouping of these junctions in order to build the high-level contour primitives, the generic volumetric structures. Experimental results for various images of clutte red scenes show an ability to properly extract the structures of volumetric objects or parts with planar and curved surfaces.
Skeletonization of two-dimesional regions using hybrid method
(Václav Skala - UNION Agency, 2007) Merad, D.; Mallem, M.; Lelandais, S.; Skala, Václav
Conversion of two-dimensional objects into a skeletal representation forms an essential step in many image processing and pattern recognition applications. Most of the topological structure of objects, and the information contained in the outline of their shapes, are preserved in the skeleton. Approaches based on Voronoi techniques preserve topology, but heuristic measures are introduced to remove unwanted edges. Methods based on Euclidean distance functions can localize skeletal points accurately, but often at the cost of altering the topology of the object. In this paper we offer a method to generate skeletal representations combining this two methods , which is robust and accurate, and preserves topology.
Vrgaments - relocatability as an extension to programmable rasterization hardware
(Václav Skala - UNION Agency, 2006) Diepstraten, Joachim; Weiskopf, Daniel; Kraus, Daniel; Ertl, Thomas; Skala, Václav
We propose an extension to the pixel pipeline of current programmable rasterization hardware to include the possibility to freely locate the rasterization position of fragments and pixels. The corresponding new primitive name is Vragment (variable fragment). We show how this new functionality could lead to new and wider classes of algorithms in computer graphics, especially in image processing, scientific visualization, geometric modeling, and rendering. A GPU-assisted simulator for programs running on the proposed architecture is presented.
Algorithms and hardware for data compression in point rendering applications
(Václav Skala - UNION Agency, 2005) Mallón, P. N.; Bóo, M.; Amor, M.; Bruguera, J. D.; Skala, Václav
The high storage requirements associated with point rendering applications make the utilization of data compression techniques interesting. Point rendering has been proposed only recently and, consequently, no compression strategies have yet been developed. In this paper we present compression algorithms for two specific data distributions widely used in point rendering: a naive distribution with no specific sorting of the points and a layer distribution which is suitable for incremental algorithms. In this last case points are sorted in layers and the connectivity among them is encoded. The algorithms we propose have a high compression rate (5.0 bits/point for the naive distribution and 7.7 bits/point for the layer distribution). Additionally we present the hardware implementation for the decompression of both algorithms. Both algorithms are implemented in a single hardware unit providing a control to select between them.
Interactive museum exhibit using pointing gesture recognition
(Václav Skala - UNION Agency, 2004) Malerczyk, Cornelius; Skala, Václav
This paper describes a Mixed Reality-supported interactive museum exhibit. Using an easy and intuitive pointing gesture recognition system, the museum visitor is able to create his/her own exhibit choosing between different painters, artistic topics or just between different images. The usage of a video-based gesture tracking system ensures a seamless integration of Mixed Reality technologies into the environment of a traditional museum. Furthermore, it addresses even technically unversed users due to the fact that no physical devices need to be used and even no training phase is necessary for the interaction. The display of digitised paintings on an interactive screen is usable e.g. in museums that do not have the space to present all of their paintings in the traditional ways. Furthermore, direct interaction with art pieces leads to a deeper involvement with and understanding of the art pieces, whereas the manipulation of original paintings is obviously prohibited. Exploration of the paintings is achieved by giving the user the possibility of looking at details of paintings, which he/she normally can only see using tools like a magnifying glass.
Visitor identification - elaborating real time face recognition system
(Václav Skala - UNION Agency, 2004) Kukharev, Georgy; Nowosielski, Adam; Skala, Václav
In this paper requirements and conditions for the visitor identification system are outlined and an example system is proposed. Two main subsystems: face detection and face recognition are described. Algorithm for face detection integrates skin-colour, mask analysis, facial features (fast and effective way of eyes localization is presented), reductors, knowledge and template matching. For face recognition a three stage algorithm is proposed. It utilizes well known methods connected in a sequential mode. To improve accuracy and speed some modifications to original methods were proposed and new one presented. The aim was to build a visitor identification system which would be able to operate in mode with a camera and present results in real-time. The emphasis on speed and accuracy was stressed.

Browse

Recent Submissions