[ Up ] [ Proposal ] [ Current Status ]

Structured Access to Information Status Report, Q2 1999

Computer Vision - Daniel Huttenlocher and Ramin Zabih

The Computer Vision Group at Cornell has received and installed 10 300MHz Pentium-II systems (4 dual processor and 6 single processor). Each of these systems is running Windows NT 4.0, and is equipped with a 21" color monitor and a Videum video capture card with a camera. Two of the sytems have pan-tilt cameras attached to the video cards. Most of these machines are installed in the Vision Lab, which is part of a larger facility including researchers in multimedia systems, networking and operating systems. There is considerable cross-group work in this facility, such as the Lecture Browser project being pursued jointly by Prof. Brian Smith's multimedia group and our group (there are slides from a talk describing this project).

Over the past 18 months, as a result of the TechEd grant and prior Intel equipment donations, we have moved our systems development to Intel Architecture machines, under Windows NT, using Visual Studio with Purify and Quantify as our development environment. We have ported our shared image libraries, as well as a number of applications that previously ran on Unix systems. All new development in the group is taking place in this environment. Completed projects since the beginning of the TechEd 2000 grant (Fall 1997) include:

1.  A video driver that provides applications with direct control over the buffer management for video cards that support Video For Windows, thus enabling applications that process video data to make efficient use of such hardware.
2.  A new version of our model-based object tracking system using the Hausdorff distance. This system uses an efficient, robust background motion estimation technique for improved model-based tracking results. This is especially important from low-resolution objects such as those in aerial imagery, which we are investigating as part of the DARPA VSAM project (there are slides from a talk describing progress on this project).
3. A graph-based image segmentation algorithm that measures local image variation, and thus is able to preserve fine detail in low variabililty regions, while merging high-variability regions (such as texture) into one (there is an abstract and recent paper, from CVPR-98).
4.  An image comparison technique based on properties of coarse "blobs" in images, that correspond to regions of extremal (top or bottom quantile) intensity, color and texture of the image (there is a recent paper, from the 1998 IEEE Workshop on Content-Based Access of Image and Video Databases)
5.  An efficient approximation to Markov Random Fields (MRF's) with applications to stereo matching (there is an abstract and recent paper, from CVPR-98).

For more information contact:
Prof. Dan Huttenlocher (dph@cs.cornell.edu)
Prof. Ramin Zabih (rdz@cs.cornell.edu)

Databases - Praveen Seshadri

The database research group at Cornell is building the PREDATOR database system as a vehicle for education and for research into query evaluation techniques for object-relational database systems. Over the last six months, the system development has moved to Windows NT running on Intel platforms, and we plan a public release of our code base in early summer 1998. Already, a number of universities and research laboratories are using PREDATOR to support their database research. Further, at Cornell, the system has been used in both introductory and advanced database classes (using PCs donated by Intel), with very positive feedback. On the research front, the ideas developed and demonstrated in PREDATOR are making their way into commercial database products, leading to substantial improvements in database querey processing efficiency, especially in the areas of decision support over complex multimedia types. Some specific recent accomplishments include the following:

For more information, check out the PREDATOR project web site.

Multimedia & Server Systems - Thorsten von Eicken and Brian Smith

Last modified on: 10/12/99