Israel SIGGRAPH meeting on

Israel SIGGRAPH meeting on February 18, 2000

Hall 12, Schreiber Bldg. (School of Math. Sciences)
Tel-Aviv University

Chair: Dan Gordon
Dept. of Computer Science
University of Haifa

The following program is also available in PostScript format. The PostScript file is to be printed double-sided on A4 paper, and folded into three columns with INVITATION and the digit "5" on the exposed columns. The invitation also serves as an entrance permit for your car at gate 1 of the Tel-Aviv University campus.

Time Speaker Title Abstract

8:30 REFRESHMENTS

9:00 Guillermo Savransky

Dept. of Computer Science

Technion Modeling and Rendering Escher-Like Impossible Scenes Inspired by the drawings of "impossible" objects by artists such as M.C. Escher, we describe a mathematical theory which captures some of the underlying principles of their work. Using this theory, we show how impossible three-dimensional scenes may be modeled and rendered synthetically.
Joint work with Dan Dimerman and Craig Gotsman

9:30 David Dobkin
Dept. of Computer Science

Princeton University Applied Computational Geometry -- Progress Report Computational Geometry has been a thriving research area for the past 20 years. During that time, the field has grown from a handful of researchers working on a small set of problems to a full-blown research area with multiple conferences and journals and many hundreds of researchers. The initial motivation for the field was to develop algorithms that would find application in practice in other fields.

In this talk, the history of the field will be traced through a few examples of how ideas developed as theoretical insights eventually found their way into quite practical situations.

10:30 COFFEE BREAK

11:00 Vladlen Koltun

Dept. of Computer Science

Tel-Aviv University Virtual Occluders: a From-Region Visibility Techniques In this talk I'll introduce the notion of virtual occluders. Given a scene and a viewcell, a virtual occluder is a view-dependent (simple) convex object, which is guaranteed to be fully occluded from any given point within the viewcell and which serves as an effective occluder from the given viewcell. Virtual occluders compactly represent the aggregate occlusion for a given cell. The introduction of such view-dependent virtual occluders enables applying an effective region-to-region or cell-to-cell culling technique and efficiently computing the PVS from a cell/region. I'll present a technique that synthesizes such virtual occluders by aggregating the visibility of a set of individual occluders and I'll show their effectiveness.
Joint work with Yiorgos Chrystanthou and Daniel Cohen-Or .

11:30 Iddo Drori

Inst. of Computer Science

The Hebrew University Image Operations in the Wavelet Domain A wide class of operations on images can be performed directly in the wavelet domain, i.e. by operating on the coefficients of the wavelet transforms of the images and/or other matrices defined by the operation. Performing such operations in the wavelet domain and then reconstructing the result is often more efficient than performing the same operation in the standard direct fashion. Perhaps more importantly, operating in the wavelet domain enables to perform these operations progressively, to operate on different resolutions, manipulate features at different scales, and to localize the operation in both the spatial and the frequency domains. In this talk we will demonstrate the applicability and the advantages of this approach to two common types of image operations: 3D image warping and convolution of images and image sequences.
Joint work with Dani Lischinski

12:00 Jacob Barhak
Dept. of Mechanical Engineering

Technion Parameterization and Surface Fitting of Range Images In reverse engineering, laser scanners are commonly used since they can sample 3D data fast and very accurately relative to other systems. However, a laser scanner system provides an enormous amount of digitized point data that is irregular and scattered and requires intensive processing in order to reconstruct the surface of the object. Reconstruction of freeform objects is based on two main stages: (1) parameterization and (2) surface fitting. Selection of an appropriate parameterization is essential for topology reconstruction as well as for surface fitness. Current parameterization methods have topological problems, thus leading to undesired surface fitting results such as noisy self-intersecting surfaces. This is particularly common for concave shapes where the parametric grid is self-intersecting or curved so that the resulting fitted surface significantly twists and changes its original shape. In this lecture, PDE & Neural Network (SOM) parameterization methods that produce a parametric grid without self-intersections will be described. In addition, iterative surface fitting methods, GDA & RSEC methods, were developed and implemented.

The feasibility of the parameterization and fitting algorithms will be demonstrated on several examples using sculptured free objects.
Joint work with Anath Fischer

Time	Speaker	Title	Abstract
8:30		REFRESHMENTS
9:00	Guillermo Savransky Dept. of Computer Science Technion	Modeling and Rendering Escher-Like Impossible Scenes	Inspired by the drawings of "impossible" objects by artists such as M.C. Escher, we describe a mathematical theory which captures some of the underlying principles of their work. Using this theory, we show how impossible three-dimensional scenes may be modeled and rendered synthetically. Joint work with Dan Dimerman and Craig Gotsman
9:30	David Dobkin Dept. of Computer Science Princeton University	Applied Computational Geometry -- Progress Report	Computational Geometry has been a thriving research area for the past 20 years. During that time, the field has grown from a handful of researchers working on a small set of problems to a full-blown research area with multiple conferences and journals and many hundreds of researchers. The initial motivation for the field was to develop algorithms that would find application in practice in other fields. In this talk, the history of the field will be traced through a few examples of how ideas developed as theoretical insights eventually found their way into quite practical situations.
10:30		COFFEE BREAK
11:00	Vladlen Koltun Dept. of Computer Science Tel-Aviv University	Virtual Occluders: a From-Region Visibility Techniques	In this talk I'll introduce the notion of virtual occluders. Given a scene and a viewcell, a virtual occluder is a view-dependent (simple) convex object, which is guaranteed to be fully occluded from any given point within the viewcell and which serves as an effective occluder from the given viewcell. Virtual occluders compactly represent the aggregate occlusion for a given cell. The introduction of such view-dependent virtual occluders enables applying an effective region-to-region or cell-to-cell culling technique and efficiently computing the PVS from a cell/region. I'll present a technique that synthesizes such virtual occluders by aggregating the visibility of a set of individual occluders and I'll show their effectiveness. Joint work with Yiorgos Chrystanthou and Daniel Cohen-Or .
11:30	Iddo Drori Inst. of Computer Science The Hebrew University	Image Operations in the Wavelet Domain	A wide class of operations on images can be performed directly in the wavelet domain, i.e. by operating on the coefficients of the wavelet transforms of the images and/or other matrices defined by the operation. Performing such operations in the wavelet domain and then reconstructing the result is often more efficient than performing the same operation in the standard direct fashion. Perhaps more importantly, operating in the wavelet domain enables to perform these operations progressively, to operate on different resolutions, manipulate features at different scales, and to localize the operation in both the spatial and the frequency domains. In this talk we will demonstrate the applicability and the advantages of this approach to two common types of image operations: 3D image warping and convolution of images and image sequences. Joint work with Dani Lischinski
12:00	Jacob Barhak Dept. of Mechanical Engineering Technion	Parameterization and Surface Fitting of Range Images	In reverse engineering, laser scanners are commonly used since they can sample 3D data fast and very accurately relative to other systems. However, a laser scanner system provides an enormous amount of digitized point data that is irregular and scattered and requires intensive processing in order to reconstruct the surface of the object. Reconstruction of freeform objects is based on two main stages: (1) parameterization and (2) surface fitting. Selection of an appropriate parameterization is essential for topology reconstruction as well as for surface fitness. Current parameterization methods have topological problems, thus leading to undesired surface fitting results such as noisy self-intersecting surfaces. This is particularly common for concave shapes where the parametric grid is self-intersecting or curved so that the resulting fitted surface significantly twists and changes its original shape. In this lecture, PDE & Neural Network (SOM) parameterization methods that produce a parametric grid without self-intersections will be described. In addition, iterative surface fitting methods, GDA & RSEC methods, were developed and implemented. The feasibility of the parameterization and fitting algorithms will be demonstrated on several examples using sculptured free objects. Joint work with Anath Fischer