|
|
 |
 |
April 2004
Friday, April 2nd 3p.m.
719 Broadway, 12th Floor Large Conference Room
High Dynamic Range Video
Sing Bing Kang
Microsoft Research
Typical video footage captured using an off-the-shelf camcorder
suffers from limited dynamic range. In this talk, I will describe our
approach to generate high dynamic range (HDR) video from an image
sequence of a dynamic scene captured while rapidly varying the
exposure of each frame. Our approach consists of three parts:
automatic exposure control during capture, HDR stitching across
neighboring frames, and tonemapping for viewing. HDR stitching
requires accurately registering neighboring frames and choosing
appropriate pixels for computing the radiance map. I will show
examples for a variety of dynamic scenes, and will dwell a bit on the
specific application of a virtual walkthrough. I will also describe
how we can compensate for scene and camera movement when creating an
HDR still from a series of bracketed still photographs.
This talk is based on joint work with Matthew Uyttendaele, Simon
Winder, and Richard Szeliski, and was presented at SIGGRAPH'03.
Sing Bing Kang received his Ph.D. in robotics from CMU in 1994. He is
currently a researcher at Microsoft Corporation working on environment
modeling from images. His paper on the Complex Extended Gaussian Image
won the IEEE Computer Society Outstanding Paper award at CVPR'91. His
IEEE Transactions on Robotics and Automation paper on human-to-robot
hand mapping was awarded the 1997 King-Sun Fu Memorial Best
Transaction Paper award. Sing Bing has published about 20 refereed
journal papers and about 45 refereed conference papers, mostly on
stereo and image-based rendering. He has also co-edited two books: one
on panoramic vision (published by Springer in 2001), and another on
"Emerging Topics in Computer Vision" (to appear in May 2004).
|
|
 |
 |
|
|
March 2004
Thursday, March 4th 2pm
719 Broadway, 12th Floor Large Conference Room
Ubiquitous Animated Characters: Removing the authoring and bandwidth
bottlenecks with performance capture and MPEG-4
Dr. Eric Petajan
Founder and Chief Scientist
face2face animation, inc
The demand for high quality animated content has produced thriving
video
game and CG feature film industries. Simultaneously, the
entertainment
industry is looking for new ways to reach the consumer wherever they
are
located. Given the current wireless bandwidth limitations and Moore's
Law, low bit-rate delivery of animated content provides the best way
to
entertain the consumer on mobile devices.
The bit-rate needed to animate a typical graphical or object-based
scene
is orders of magnitude less than the compressed video bit-rate of the
rendered scene. The development of MPEG-4 animation coding is
motivated
by the need for high quality visual communication at low bit-rates
coupled
with low-cost graphics rendering systems.
MPEG-4 contains a comprehensive set of tools for representing and
compressing content objects and the animation of those objects.
Virtual
humans (faces and bodies) are treated as a special type of object in
MPEG-4 with anatomically specific locations and animation parameters
specified in the standard. While virtual humans can be treated as
generic
graphical objects, there are particular advantages to representing
them
with the Face and Body Animation (FBA) Coding specification. A facial
motion capture system has been developed which transfers the lip and
head
motion from ordinary video of a real human to an animated character
using
only the MPEG-4 Face Animation Parameters (FAPs).
The perception of human speech incorporates both acoustic and visual
communication modalities. Automatic speech recognition (ASR) systems
have
traditionally processed only the acoustic signal. While video cameras
and
video acquisition systems have become economical, the use of automatic
lipreading to enhance speech recognition performance is an ongoing and
fruitful research topic. During the last 20 years a variety of
research
systems have been developed which demonstrate that visual speech
information enhances overall recognition accuracy, especially in the
presence of acoustic noise. The performance of all of these systems
could
be greatly improved by better video feature acquisition or better
acoustic/visual recognition integration methods or both.
face2face animation has developed a facial motion capture system for
the
efficient authoring of high quality animated characters. The animation
quality is high enough for broadcast television (HBO) and the bit-rate
is
low enough for communication over mobile or dialup networks.
face2face
has recently created the first 3D talking characters on a mobile
phone. This talk will cover the MPEG-4 standard and the many options
for
delivering high quality talking characters to the consumer.
|
|
| |
|
|
|
March 2004
Thursday, March 25th, 2004. 6 pm
719 Broadway, 12th Floor Large Conference Room
Please join us for a viewing of Part I of
The Destruction of the Schwartzes
The Holocaust memoirs of Jack Schwartz's cousin Tibor Schwartz, one of
3
survivors of about 30 relatives on Jack Schwartz's father's side.
The film covers the period from 1944-1946.
This is one of the thousands of interviews taped by the Spielberg Shoah
Foundation.
|
|
| |
|
|
|
February 2004
Monday, February 9th 7:00-9:00 p.m.
NYU's Kimmel Center at 60 Washington Square
On Monday, February 9 from 7:00PM to 9:00PM, NYC ACM
SIGGRAPH is very proud to present the SIGGRAPH 2003 Electronic Theater.
This screening will be held at NYU's Kimmel Center at 60 Washington
Square South in the Village.
The SIGGRAPH 2003 Electronic Theater is part of the world's
most prestigious film and video extravaganza, showcasing dazzling
and innovative imagery in invited and submitted works selected by a
distinguished jury of computer graphics experts and specialists. As
part of the Computer Animation Festival, the SIGGRAPH 2003
Electronic Theater is internationally recognized and lauded as an event
that serves to engage and inspire artists, scientists, engineers,
designers, and students to harness the power of the digital image to
explore the boundlessness of imagination.
NYC ACM SIGGRAPH is very proud to have SIGGRAPH 2004
Computer Animation Festival (CAF) Chair Chris Bregler of NYU as our
guest host for the evening. Chris will also be sharing some insights into
how the CAF works. You can find out more information about this year's
conference, SIGGRAPH 2004, at www.siggraph.org/s2004.
As always, this event is free for all NYC ACM SIGGRAPH
members. It is also free for NYU faculty, staff and students. It is $7.00
for non-members and $3.00 for non-member students (with a current,
valid student ID). NYC ACM SIGGRAPH wishes to thank NYU for its generous
support of this event. More information about the presentation and
directions to the Kimmel Center can be found on our web site at
nyc.siggraph.org
|
|
| |
|
|
|
February 2004
Friday, February 13, 2 pm
719 Broadway, 12th Floor Large Conference Room
Ms. Page will be speaking about Shrek 2 and other new shows from
DreamWorks. She will also be showing highlights from Imagina, and
other
cool stuff from around Europe--short films and commercials, and also
some
of the particle work shown at Imagina by Bruce Glass.
|
|
| |
|
|
|
February 2004
Wednesday February 25th, 2004: 2:30 - 3:30 p.m.
719 Broadway, 12th Floor Large Conference Room
Serge J. Belongie
Department of Computer Science and Engineering
University of California, San Diego
Lecture:
Three Brown Mice: See How They Run -- Monitoring Rodent Behavior in
the
Smart Vivarium
We address the problem of tracking multiple, identical, nonrigid
moving targets through occlusion for purposes of rodent surveillance
from a side view. Automated behavior analysis of individual mice
promises to improve animal care and data collection in medical
research. In our experiments, we consider the case of three brown
mice that repeatedly occlude one another and have no stable trackable
features. Our proposed algorithm computes and incorporates a hint of
the future location of the target into layer-based affine optical flow
estimation. The hint is based on the estimated correspondences
between mice in different frames derived from a depth ordering
heuristic. Our approach is simple, efficient, and does not require a
manually constructed mouse template. We demonstrate encouraging
results on a challenging test sequence containing multiple instances
of severe occlusion. (This is joint work with Kristin Branson and
Vincent Rabaud.)
|
|
| |
|
| |
For additional information, contact: info@cat.nyu.edu
|
|
|
|
|
| |
|
|
|
 |
