Spring 2006 Series

March 20th

Richard Szeliski (Microsoft)

April 21st

Steve Marschner (Cornell)

May 19th Marc Levoy (Stanford)
May 22nd

Doug James (CMU)

Spring 2005 Series

  February 28th

  Marc Levoy (Stanford)

Fall 2004 Series

  November 18th

  Marc Raibert (Boston Dynamics)
Spring 2004 Series
  April 8th
  Dimitris Metaxas (Rutgers)
Fall 2003 Series
  November 5th
  Martial Hebert (CMU)
  December 10th   Yiannis Aloimonos (UMD)
Spring 2003 Series

  February 3rd

  Demetri Terzopoulos (NYU)
  February 19th   Takeo Kanade (CMU)
  March 5th   Alex Pentland (MIT)
  April 2nd   Jessica K. Hodgins (CMU)
  April 16th   Michael J. Black (Brown)
Fall 2002 Series
  December 4th
 Dinesh K. Pai (Rutgers)

   Doug James

 School of Computer Science, Carnegie Mellon University

 "Precomputed acoustic transfer"

 Date May 22nd, 2006
 Location  InterSchool Lab., Schapiro CEPSR, 11:00AM
 Abstract

Simulating sounds produced by realistic vibrating objects is challenging because sound radiation involves complex diffraction and interreflection effects that are very perceptible and important. These wave phenomena are well understood, but have been largely ignored in computer graphics due to the high cost and complexity of computing them at audio rates.

We describe a new algorithm for real-time synthesis of realistic sound radiation from rigid objects. We start by precomputing the linear vibration modes of an object, and then relate each mode to its sound pressure field, or acoustic transfer  function, using standard methods from numerical acoustics. Each transfer function is then approximated to a specified accuracy using low-order multipole sources placed near the object. We provide a low-memory, multilevel, randomized algorithm for optimized source placement that is suitable for complex geometries. At runtime, we can simulate new interaction sounds by quickly summing contributions from each mode’s equivalent multipole sources. We can efficiently simulate global effects such as interreflection and changes in sound due to listener location. The simulation costs can be dynamically traded-off for sound quality. We present several examples of sound generation from physically based animations.

 

 Biography

Doug L. James has been an Assistant Professor of Computer Science and Robotics at Carnegie Mellon University since Fall 2002. He received his Ph.D. from the Institute of Applied Precomputed Acoustic Transfer Mathematics at the University of British Columbia, Vancouver, Canada, advised by Dinesh K. Pai. Doug received an NSF Early Career Development Award for his work on “Precomputing Data-driven Deformable Systems for Multimodal Interactive Simulation,” he was one of Popular Science magazine’s “Brilliant 10” young scientists for 2005, and is an Alfred P. Sloan research fellow.

back to top

 

Marc Levoy
   Marc Levoy

 Computer Science Department, Stanford University

 "Light field microscopy"

 Date May 19th, 2006
 Location  InterSchool Lab., Schapiro CEPSR, 11:00AM
 Abstract

By inserting a microlens array into the optical train of a conventional microscope, light fields can be captured of tiny biological (or other) specimens in a single snapshot. Although diffraction limits the product of spatial and angular resolution in these light fields, we can nevertheless produce useful perspective flyarounds and 3D focal stacks from them. Since microscopes are inherently orthographic devices, perspective flyarounds represent a new way to look at microscopic specimens. Focal stacks are not new, but manual techniques for capturing them are time-consuming and hence not applicable to moving or light-sensitive specimens. Applying 3D deconvolution to these focal stacks, we can produce a set of cross sections, which can be visualized using volume rendering. Ours is the first technology (of which we are aware) that can produce such a volumetric model from a single photograph, albeit only of microscopic objects.

In this talk, I will describe a prototype light field microscope, analyze its optical performance, and show perspective views, focal stacks, and reconstructed volumes for a variety of biological specimens. I will also show that synthetic focusing followed by 3D deconvolution is equivalent to applying limited-angle tomography directly to the 4D light field. Simply stated, these two techniques, which developed in different scientific disciplines, turn out to be the same thing.

 

 Biography

Marc Levoy is a Professor of Computer Science and Electrical Engineering at Stanford University. He received degrees in Architecture from Cornell University in 1976 and 1978 and a PhD in Computer Science from the University of North Carolina in 1989. His research interests include computer-assisted cartoon animation, volume rendering (for which he won the SIGGRAPH Computer Graphics Achievement Award in 1996), 3D scanning, light field sensing and display, computational imaging, and digital photography.
 

back to top

 

   Steve Marschner

 Computer Science Department, Cornell University

 "Scattering and multiple scattering in fibrous materials"

 Date April 21st, 2006
 Location  InterSchool Lab., Schapiro CEPSR, 11:00AM
 Abstract

Some very common materials -- hair, fur, feathers, wood, fabric -- are constructed out of thin fibers, often in highly organized structures. Because of their geometric structure, many of these materials have unusual optical properties that don't fit the standard assumptions about directional reflectance and scattering that we like  to make in graphics and vision. I'll discuss several research projects that address modeling and simulation of illumination for materials made of fibers. The first is a geometric analysis of light scattering from individual fibers of human hair. The second is a model for light reflection from wood surfaces like furniture or floors. Because boards are in general cut at an angle to the tree's growth direction, the subsurface reflection in wood is often quite asymmetric. The third and most recent is a project developing algorithms for light transport simulation in hair. Especially in light-colored hair, multiple scattering is essential to accurate lighting. We argue that the problem should be approached volumetrically: a mass of hair fibers acts like an anisotropic and highly orientation-dependent scattering medium.

 

 Biography

Steve Marschner joined the faculty at Cornell in Fall 2002; my research and teaching center around computer graphics. Most of his research is carried out in the Program of Computer Graphics. Before joining to Cornell he worked in the Computer Graphics lab at Stanford on material appearance modeling and on the The Digital Michelangelo Project, and before that at Microsoft Research and at HP Laboratories. His Ph.D. is from Cornell in 1998.

back to top

 

   Richard Szeliski

 Interactive Visual Media Group, Microsoft Research

 "Why matting matters"

 Date March 20th, 2006
 Location  InterSchool Lab., Schapiro CEPSR, 2:00PM
 Abstract

Image matting (e.g., blue-screen matting) has been a mainstay of Hollywood and the visual effects industry for decades, but its relevance to computer vision is not yet fully appreciated. In this talk, I argue that the mixing of pixel color values at the boundaries of objects (or even albedo changes) is a fundamental process that must be correctly modeled to make meaningful signal-level inferences about the visual world, as well as to support high-quality imaging transformations such as de-noising and de-blurring. Starting with Ted Adelson et al.'s seminal work on layered motion models, I review early stereo matching algorithms with transparency and matting (with Polina Golland), work on layered representations with matting (with Simon Baker and Anandan), through Larry Zitnick's 2-layer representation for 3D video. I then present our recent work (with Ce Liu et al.) on image de-noising using a segmented description of the image and Eric Bennett et al.'s work on multi-image de-mosaicing, again using a local two-color model.

 

 Biography

Richard Szeliski leads the Interactive Visual Media Group at Microsoft Research, which does research in digital and computational photography, video scene analysis, 3-D computer vision, and image-based rendering. He received a Ph.D. degree in Computer  Science from Carnegie Mellon University, Pittsburgh, in 1988. He joined Microsoft Research in 1995. Prior to Microsoft, he worked at Bell-Northern Research,  Schlumberger Palo Alto Research, the Artificial Intelligence Center of SRI International, and the Cambridge Research Lab of Digital Equipment Corporation.

Dr. Szeliski has published over 100 research papers in computer vision, computer graphics, medical imaging, and neural nets, as well as the book Bayesian Modeling of Uncertainty in Low-Level Vision. He was a Program Committee Chair for ICCV'2001 and the 1999 Vision Algorithms Workshop, served as an Associate Editor of the IEEE Transactions on Pattern Analysis and Machine Intelligence and on the Editorial Board of the International Journal of Computer Vision, and is a Founding Editor of Foundations and Trends in Computer Graphics and Vision.

back to top


Marc Levoy
   Marc Levoy

 Computer Science Department, Stanford University

 "Light field photography and videography"

 Date February 28th, 2005
 Location  InterSchool Lab., Schapiro CEPSR, 4:00PM
 Abstract

The light field is a 4D representation of radiance as a function of position and direction in space. In computer graphics, light fields have been used to fly through scenes without the use of geometric models. In this talk, I explore three ways to capture and use light fields that fall outside this paradigm. Specifically, I describe:
1. A new photographic technique called dual photography, which exploits Helmholtz reciprocity to interchange the lights and cameras in a scene. In its simplest form, the technique allows us to take photographs using a projector and a photocell. Replacing the photocell with a camera or an array of cameras produces a 4D or 6D dataset, with applications to relighting and the measurement of appearance.
2. A compact handheld camera capable of capturing a light field in a single exposure. The main idea is to insert a microlens array between the sensor and main lens. By capturing directional as well as spatial information about the light entering the camera, we can refocus a photograph *after* it is taken, and we can move the viewpoint slightly.
3. New applications for the Stanford multi-camera array. In past work, we have configured our array to generate video at 3000 frames per second and to simulate a camera with an 8-foot aperture - allowing us to see through foliage and crowds. Recently, we have configured the array to simulate a 30-megapixel tiled video camera with independent exposure metering in each tile. This lets us record dynamic environments with unprecedented resolution and dynamic range.

 

 Biography

Marc Levoy is an Associate Professor of Computer Science and Electrical Engineering at Stanford University. He received his PhD in Computer Science from the University of North Carolina in 1989. In the 1970's Levoy worked on computer animation, developing an early computer-assisted cartoon animation system. In the 1980's Levoy worked on volume rendering, a family of techniques for displaying sampled three-dimensional functions, such as CT and MR data. In the 1990's he worked on technology and algorithms for 3D scanning. This led to the Digital Michelangelo Project, in which he and a team of researchers spent a year in Italy digitizing the statues of Michelangelo using laser rangefinders. His current interests include light field sensing and display, computational imaging, and digital photography. Levoy received the NSF Presidential Young Investigator Award in 1991 and the SIGGRAPH Computer Graphics Achievement Award in 1996 for his work in volume rendering.

back to top

Marc Raibert
   Marc Raibert

 Boston Dynamics

 "Dynamic Legged Robots: Past, Present and Future"

 Date November 18th, 2004
 Location  InterSchool Lab., Schapiro CEPSR, 11:00 AM
 Abstract

Twenty five years ago the Leg Laboratory set out to create a family of dynamic legged robots -- robots that would run, jump, balance actively and do gymnastic maneuvers. Last year we revived the LegLab with the goal of creating a new generation of robots that will travel on rough terrain, terrain too steep, rugged and varied for wheeled or tracked vehicles. In this talk I will review the old work, describe progress on the legged robots we are building today, and say a few words about where we hope legged robots will take us in the future.

 

 Biography

Marc Raibert, founder of Boston Dynamics, was formerly Professor of Electrical Engineering and Computer Science at MIT. In previous work, he developed laboratory robots that used control systems for balance and to coordinate their motions. These robots provided new insights into the science and art of how people move and inspired the launch of Boston Dynamics in 1992, a software engineering company specializing in human simulation. They create the full spectrum of human simulations used for a wide range of applications, from training and mission planning to biomechanical analysis and virtual prototyping.

back to top

Dimitris Metaxas
   Dimitris Metaxas

 Dept. of Computer Sciences, Rutgers University

 "Hybrid Deformable Modeling Methods"

 Date April 8th, 2004
 Location  InterSchool Lab., Schapiro CEPSR, 11:00 AM
 Abstract

Recent advances in deformable models have lead to new classes of methods that borrow the best features form level sets as well as traditional parametric deformable models. In this talk I will first present a new class of such models termed Metamorphs whose formulation integrates shape, intensity and texture by borrowing ideas from level set and traditional parametric deformable models. These new models can be used in medical segmentation and registration where organ boundaries are fuzzy and with no assumptions on the noise distribution. In the second part of the talk I will present novel body and face tracking methods based on a novel use of deformable models and registration methods. Time permitting I will show some new fluid and cloth animation results.

 

 Biography

Dr. Dimitris Metaxas is a Professor in the Division of Computer and Information Sciences and Professor in the Department of Biomedical Engineering at Rutgers University since September 2001. He is directing the Center for Computational Biomedicine, Imaging and Modeling (CBIM). From January 1998 to September 2001 he was a tenured Associate Professor in the Computer and Information Science Department of the University of Pennsylvania and Director of the VAST Lab. Prior to this he was an Assistant Professor in the same department since 1992. Prof. Metaxas received a Diploma in Electrical Engineering with highest honors from the National Technical University of Athens Greece in 1986, an M.Sc. in Computer Science from the University of Maryland, College Park in 1988, and a Ph.D. inComputer Science from the University of Toronto, Ontario, Canada in 1992.

back to top

Martial Hebert
   Martial Hebert

 Robotics Institute, Carnegie Mellon University

 "Experiments in Object Recognition"

 Date November 5th, 2003
 Location  InterSchool Lab., Schapiro CEPSR, 11:00 AM
 Abstract

In this talk, I will describe research that we have been conducted over the past couple of years in the area of object recognition. The goal of this work is to recognize objects, or classes of objects in images with clutter and under varying viewpoint and illumination conditions. I will start with the "simple" case of recognizing specific objects that can be characterized by collections of image patches. I will then address the more difficult problem of recognizing objects that cannot be described by local distribution of pixel values. For such objects, we need to use a description of the shape rather than of the local distribution of pixel values. In turns out that this is difficult to do when one wants to be robust to background clutter and to have good generalization to changes in the object shape. Finally, I will say a few words about recent work on a much more difficult problem: The recognition of large classes of objects. This includes, for example, recognizing man-made structures in images or recognizing deformable objects. For that, we have recently introduced a new approach based on random field techniques that address a number of open issues in this area. I will illustrate the limitations and challenges in recognition by using the results obtained in each of these areas.

 

 Biography

Martial Hebert is professor at the Robotics Institute, Carnegie Mellon University. His interests include object recognition from images, image sequences, and 3-D range data; the automatic generation of 3-D models for large sets of unregistered frames; and environment understanding from sensor data for autonomous driving for mobile robots. In the area of recognition, his group has developed a variety of techniques for recognizing objects and classes of objects in image and range data out of large databases of object models. In the area of mobile robotics, his group has developed techniques for automatic terrain classification from sensor data which have been used on several different mobile plaftforms and with different sensors.

back to top

Yiannis Aloimonos
   Yiannis Aloimonos

 Center for Automation Research, University of Maryland

 "Vision is a Chicken Egg Problem:
 The Correspondence Problem"

 Date December 10th, 2003
 Location  InterSchool Lab., Schapiro CEPSR, 11:00 AM
 Abstract

I will start with visual illusions. I will show several of them and I will give you a simple computational principle that explains (predicts) almost all geometric optical illusions. It turns out there are inherent limitations in estimating a point, a line or movement in images, that is in finding features. Yet, if we concentrate on the processes responsible for 3D vision, the state of the art theory of multiview geometry is doing geometry on points, lines and movement. As a result, the outputs of our best programs are not good enough for many tasks.The conventional wisdom is that if could correspond or match images, then the problem is solved. But Vision is a compositional process; it does not consist of a clearly defined set of steps of a feedforward nature. It is rather a set of interacting processes existing in a set of feedback loops, and correspondence is one of these processes. Hence, the chicken-egg aspect arises. We need correspondence to get to 3D shape and movement, but we need the 3D stuff to better estimate correspondence. I will describe our work on Feedback Vision and provide examples from stereo, segmentation and motion analysis. It turns out that to do feedback vision, we have to use the signal, i.e. its harmonic components, and do geometry with them. This sort of Harmonic Computational Geometry, which does Geometry on the output of filters applied to images, has the potential to considerably advance the field. In general, bridging the gap between signals and geometry, is an open and exciting discipline.

 

 Biography

Yiannis Aloimonos studied Mathematics in Athens, Greece and Computer Science at the University of Rochester (PhD 1987). He is a Professor in the Computer Science Department of the University of Maryland, and the Director of the Computer Vision Laboratory at the Institute for Advanced Computer Studies. He is interested in the relationship of action to intelligence and the mathematics of a sensorimotor theory of cognition. For the past twenty years he has been working on 3D vision (multiview geometry, trilinear constraints, active vision, video processing). Recent work from his group is featured in the New Scientist. See also a Reuters report.

back to top

Demetri Terzopoulos
   Demetri Terzopoulos

 Courant Institute, New York University

 "Physical and Biological Modeling for Graphics and Vision"

 Date February 3rd, 2003
 Location  InterSchool Lab., Schapiro CEPSR, 11:00 AM
 Abstract

Physics-based modeling has revolutionized computer graphics and computer vision since the mid 1980s. In this talk, I will first review deformable models and illustrate some of the many ways in which this powerful family of physics-based modeling primitives has enabled us to close the vision/graphics gap. Next, I will turn to biological modeling, an exciting new paradigm which in recent years has taken root in both domains. In this context, we have created realistic virtual worlds inhabited by "artificial animals". These lifelike synthetic fauna possess biomechanically simulated bodies, sensors, and brains with motor, perception, and behavior centers. As biomimetic autonomous agents situated in physics-based virtual worlds, artificial humans and lower animals also foster a computationally oriented understanding of biological information processing, including vision, learning, and cognition. Finally, I will show how recent work in medical image analysis, an important application area of computer vision, attempts to instill some artificial life into deformable models, leading to automatic image segmentation algorithms called "deformable organisms".

 

 Biography

Demetri Terzopoulos holds the Lucy and Henry Moses Professorship in Science at New York University and is Professor of Computer Science and Mathematics at NYU's Courant Institute. He graduated from McGill University and obtained the PhD in EECS from the Massachusetts Institute of Technology (MIT). His published work comprises hundreds of research papers and several volumes in computer graphics, computer vision, medical imaging, computer-aided design, artificial intelligence, and artificial life. Professor Terzopoulos is a Fellow of the IEEE. His research has received awards from the IEEE, American Association for Artificial Intelligence, International Digital Media Foundation, Ars Electronica, Canadian Image Processing and Pattern Recognition Society, and other organizations. He joined NYU from the University of Toronto where he is Professor of Computer Science and Professor of Electrical and Computer Engineering (on leave) and where he received six Excellence Awards and has held Canada's most prestigious fellowships in science and engineering.

back to top

Takeo Kanade
   Takeo Kanade

 Robotics Institute, Carnegie Mellon University

 "Subspace Methods for Image Analysis"

 Date February 19th, 2003
 Location  InterSchool Lab., Schapiro CEPSR, 11:00 AM
 Abstract

In 1990, Tomasi and Kanade developed a new solution method, named a factorization method, for the structure-from-motion problem. The factorization method is based on a simple observation that when the trajectories of features in a sequence of video are organized as a matrix where the entries are the feature's image coordinates with the row corresponding to the frame and the column to the feature number, the rank of the matrix is, surprisingly, only up to 3. This linear algebraic property strongly constrains the solution space, and thus allows for a stable simultaneous recovery of motion and shape. Since then, many researchers found similar subspace constraints and exploited them in solving not only structure from motion problems but other vision problems. They include multi-body motion segmentation, optical flow analysis, non-rigid shape recovery, color analysis, and even force sensor calibration.
I will first review these subspace-based methods, and discuss the common thread that has brought about the power of their solution method - the subspace constraints that come from the very physical or geometrical nature of the problem itself, rather than coming from the statistical distribution of the dataset, such as PCA, though both share the same technique. Then I will present a recently developed robust method of layer extraction from video. It is again based on the property that the relative (rather than absolute) affine transform of layer regions form an up-to-3 subspace. This region-based subspace approach, as opposed to previous point-feature based ones, allows us to robustly identify the number of layers contained in the scene including those moving, and to segment the image into layers.

 

 Biography

Takeo Kanade is U. A. Helen Whitaker University Professor of Computer Science and Robotics at Carnegie Mellon University. He received his Doctoral degree in Electrical Engineering from Kyoto University, Japan, in 1974. After holding a faculty position at Department of Information Science, Kyoto University, he joined Carnegie Mellon University in 1980, where he was the Director of the Robotics Institute from 1992 to 2001.
Dr. Kanade has worked in multiple areas of robotics: computer vision, multi-media, manipulators, autonomous mobile robots, and sensors. He has written more than 250 technical papers and reports in these areas, as well as more than 15 patents. He has been the principal investigator of more than a dozen major vision and robotics projects at Carnegie Mellon.
Dr. Kanade has been elected to the National Academy of Engineering. He is a Fellow of the IEEE, the ACM, and American Association of Artificial Intelligence (AAAI), and the former and founding editor of International Journal of Computer Vision.

back to top

Alex Pentland
   Alex Pentland

 Media Lab., Massachusetts Institute of Technology

 "Human Design: Building Computation Around Human Networks"

 Date March 5th, 2003
 Location  InterSchool Lab., Schapiro CEPSR, 11:00 AM
 Abstract

I will argue that an active computer analysis of interactions within the workplace can dramatically improve the functioning of organizations by dynamically adapting communications networks to suit the participant's behavior. There are several different types of information inherent in workplace that are accessible to machine perception: vocal prosidy and body language, participant identity, situation and location, and speech content. By using wearable devices to measure this information, and stochastic graphical modeling techniques to describe and predict the dynamics of the interactions, high-potential collaborations and expertise within the organization can be identified. Initial results at modeling human-centered networks from machine perception (including vocal prosidy, speech recognition and machine vision), and using these models to initiate productive connections will be shown, and privacy concerns addressed.

 

 Biography

Prof. Alex (Sandy) Pentland is a pioneer in wearable computers, health systems, smart environments, and technology for developing countries. He is one of the most-cited computer scientists in the world.
He is the founding director of the Media Lab Asia, and is a co-founder of the Center for Future Health, the Wearable Computing research community, and the international Digital Nations Consortium. He was formerly the Academic Head of the MIT Media Laboratory, and is the Toshiba Professor of Media Arts and Sciences.
He has won numerous international awards in the Arts, Sciences and Engineering. He was chosen by Newsweek as one of the 100 Americans most likely to shape the next century.
He currently directs the Human Design research group at the MIT Media Lab.

back to top

Jessica K. Hodgins
   Jessica K. Hodgins

 Robotics Institute, Carnegie Mellon University

 "Data Driven Animation of Human Characters"

 Date April 2nd, 2003
 Location  InterSchool Lab., Schapiro CEPSR, 11:00 AM
 Abstract

Motion capture data can be viewed as an instantiation of the motor tape hypothesis from the motor control literature. With this model, sensor or control signals are assumed to be stored as functions of time and played back to perform the desired action. In this talk, I will explore three projects in which we are looking at ways to address the major criticisms of the motor tape hypothesis: the size of the required database and the ability to respond to disturbances. In the first, we used a database to construct a graph structure that can be searched to find appropriate motion based on user input. In the second, we demonstrated that motion data can be generalized to allow a variety of manipulation tasks to be performed based on a small database. In the final project, we combined motion data with simulation to create realistic collisions for boxing.

 

 Biography

Jessica Hodgins joined the Robotics Institute and Computer Science Department at Carnegie Mellon University as a Associate Professor in fall of 2000. Prior to moving to CMU, she was an an Associate Professor and Assistant Dean in the College of Computing at Georgia Institute of Technology. She received her Ph.D. in Computer Science from Carnegie Mellon University in 1989. Her research focuses on computer graphics, animation, and robotics. She has received a NSF Young Investigator Award, a Packard Fellowship, and a Sloan Fellowship. She was editor-in-chief of ACM aTransactions on Graphics from 2000-2002 and will be SIGGRAPH Papers Chair in 2003.

back to top

Michael J. Black
   Michael J. Black

 Dept. of Computer Science, Brown University

 "The Probabilistic Inference of 3D Human Motion"

 Date April 16th, 2003
 Location  InterSchool Lab., Schapiro CEPSR, 11:00 AM
 Abstract

A probabilistic method for tracking 3D articulated human figures in monocular image sequences is presented. Within a Bayesian probabilistic framework we learn statistical models of objects and scenes and exploit these models for tracking complex, deformable, or articulated objects in image sequences. In particular, we learn the likelihood of observing various spatial and temporal filter responses corresponding to edges, ridges, and motion differences given a model of a person. Similarly, we learn probability distributions over filter responses for general scenes that define a likelihood of observing the filter responses for arbitrary backgrounds. We then derive a probabilistic model for tracking that exploits the ratio between the likelihood that image pixels corresponding to the foreground (person) were generated by an actual person or by some unknown background. A prior probability distribution over possible human motions is learned from 3D motion-capture data and is combined with the likelihood for Bayesian tracking using particle filtering. This prior term exploits ideas from texture synthesis to construct implicit probabilistic models of human motion that replace the problem of representation with that of efficient search. By combining multiple image cues, by using learned likelihood models, and by using learned prior models of motion, we demonstrate the tracking of people in monocular image sequences with cluttered scenes and a moving camera.
Joint work with:
Hedvig Sidenbladh (Swedish Defense Research Agency (FOI))
David Fleet (PARC Inc.)
Leonid Sigal (Brown University)

 

 Biography

Michael Black received his Ph.D. in computer science from Yale University in 1992. He has been a visiting researcher at the NASA Ames Research Center and an Assistant Professor in the Dept. of Computer Science at the University of Toronto. In 1993 Prof. Black joined the Xerox Palo Alto Research Center where he managed the Image Understanding area and later founded the Digital Video Analysis group. In 2000, Prof. Black joined the faculty of Brown University where he is currently Associate Professor of Computer Science. At CVPR'91 he received the IEEE Computer Society Outstanding Paper Award for his work with P. Anandan on robust optical flow estimation. In 1999 his paper with David Fleet on probabilistic detection and tracking of motion discontinuities received Honorable Mention for the Marr Prize. Prof. Black's research interests include optical flow estimation, human motion analysis, robust statistics, and brain-computer interfaces.

back to top

Dinesh K. Pai
   Dinesh K. Pai

 Div. of Computer and Information Sciences, Rutgers University

 "Multisensory Stimulation and Interaction"

 Date December 4th, 2003
 Location  InterSchool Lab., Schapiro CEPSR, 11:00 AM
 Abstract

Humans experience the world with all their senses, including vision, touch, and hearing. Therefore, human interfaces benefit from providing correlated multisensory information. I will describe recent progress in my group towards constructing multisensory environments with integrated graphics, haptics, and sounds. These environments are designed to be both realistic and responsive to human interaction. I will describe how we can construct physically based models suitable for multisensory interactive simulation, and how these models could be acquired using the UBC Active Measurement Facility (ACME). I will also describe the new Rutgers Haptic, Auditory, and Visual Environment, currently under development.

 

 Biography

Dinesh K. Pai is a Professor in the Department of Computer Science at Rutgers, the State University of NJ. He is also a Professor at the University of British Columbia and a fellow of the BC Advanced Systems Institute. He received his Ph.D. from Cornell University, Ithaca, NY. His research interests span the areas of graphics, robotics, and multisensory human-computer interaction. One current research focus is reality-based modeling, i.e., building multisensory computational models of the physical world from measurements. This includes a recent thrust in reconstruction from medical ultrasound images. Another focus is fast simulation with integrated sound, haptics, and graphics, especially simulation of contact.

back to top