Scott Cohen
[Scott Cohen]


I received my undergraduate B.S. degrees in Mathematics and Computer Science from Stanford University in June 1993. The next year I enrolled in the Ph.D. program in Computer Science at Stanford. I received my M.S. in Computer Science from Stanford in January 1996, and my Ph.D. in Computer Science from Stanford in June 1999. My advisor was Professor Leonidas Guibas. I spent 10 years at Stanford (!), 4 years as an undergrad and 6 years as a graduate student.

When I started the Ph.D. program I was heavily involved in scientific computation. I spent the first two summers of the program working at Lawrence Livermore National Laboratory on the CVODE ordinary differential equation solver. Since then, my interests have shifted to computer vision. I was a member of the Stanford Vision Laboratory (SVL) in the Image Retrieval Systems Group (IRS) group. My research focused on building content-based image retrieval systems that find color and shape query patterns within database images.

I will be working in the Graphics Technology Group at Adobe starting July 1999.


Brief descriptions of my Stanford research projects are given below.

SEDL    Finding Color and Shape Patterns in Images
The goal of this project is a system which efficiently retrieves database images that contain a given query pattern or a region similar to the query. The proposed solution is the SEDL (Scale Estimation for Directed Location) image retrieval system. The SEDL framework is general enough to find both color and shape patterns. It has been successfully applied to find (1) color patterns in a database of product advertisements with product logos as queries, and (2) shape patterns in a database of Chinese characters.
EMDG    Computing the Earth Mover's Distance under Transformations
This project aims at extending the Earth Mover's Distance (EMD) between distributions to be invariant to some given set of distribution transformations. This problem is motivated by examples from content-based image retrieval. A monotonically convergent iteration is given, although the iteration may converge to only a locally optimal transformation. Specific cases are identified in which a globally optimal transformation can be computed directly, without the aid of our iteration. A gentle introduction to the EMD is also included.

Click on a project logo for more information.


Other Works

Stanford Stanford Computer Science Department Stanford Robotics Lab Stanford Vision Lab
Page maintained by