I work with James Davis on Markerless Motion Capture of people.
We've also worked on other projects in Computer Vision and Computational Photography.
I had a fun course project, Recording a Game of Go, for James and Manfred Warmuth,
and another fun one, displaying Flickr pictures in a photo frame to match a room's decor, for Yi Zhang
A pdf of my resume is here.
My LinkedIn page is here.
I'm a Ph.D. student in computer science at the University of California, Santa Cruz, probably graduating in 2011.
Lately I've been rock climbing at Pacific Edge, playing
go at
Pergolesi, and playing ultimate frisbee at De Laveaga Park.
I worked for a few years at General Atomics
in San Diego, making Synthetic Aperture Radars (a kind of camera) on remote-control airplanes estimate where the stuff in their pictures are.
I studied mechanical engineering as an undergrad at
Princeton.
As a graduate student, I've interned at Microsoft Research,
National Semiconductor, and
Los Alamos National Labs.
As an undergrad, I interned at the Salk Institute's Computational Neurobiology Lab and the
Laboratoire National des Ponts et Chaussees.
Optimized Image Sampling for View and Light Interpolation
Prabath Gunawardane, Oliver Wang, Steve Scher, James Davis, Ian Rickard, Tom Malzbender.
International Symposium on Virtual Reality, Archaeology and Cultural Heritage (VAST) 2009.
Joint View-Interpolation and Lighting-Interpolation in image-based rendering benefits from a "View First, Lighting Second" order, the opposite of the standard approach.
The optimal tradeoff between the number of viewpoints and the number of lighting conditions at each viewpoint is determined,
as well as the optimal lighting model. Together, these methods reduce the number of photos needed.
BibTex
Material Classification with BRDF Slices
Oliver Wang, Prabath Gunawardane, Steven Scher, and James Davis.
Computer Vision and Pattern Recognition (CVPR) 2009.
Photographed an object lit from many directions.
A BRDF-based similarity measure (SVM kernel) performs better pixelwise classification of material types than the standard euclidean distance.
Paper PDF , BibTex
Making The Real World Virtual: Tracking Board Game Pieces
Steven Scher, Ryan Crabb, and James Davis.
International Conference on Pattern Recognition (ICPR) 2008.
Determined the most likely legal sequence of moves in a board game (Go) from a video of the game, using a Hidden Markov Model.
Board detected with RANSAC, pieces detected with SVM.
Paper PDF ,
Poster PDF ,
Source Code,
BibTex
High Resolution Stereoscopic SAR and Relative GPS for Time Critical Targeting
Keith Beals, Robin Snider, Steven Scher, Matt Barre, and Michael Funk
National Fire Control Symposium (NFCS) 2002.
Empirical tests of the system described in the "Precision SAR Targeting" paper below
BibTex
Precision SAR Targeting using Stereoscopy and RGPS aboard an UAV
Keith Beals, Robin Snider, Jim Broesch, Matt Barre, Steven Scher, and Michael Funk.
1st UAV, Systems, Tech, and Operations Conference and Workshop, 2002
An Unmanned Air Vehicle (UAV) and Synthetic Aperture Radar (SAR) were used to obtain 3D models and precise locations of a scene in near real time
Paper PDF ,
BibTex
Contact me at:
sscher {at} soe.ucsc.edu
Page last modified 1/6/2009