Michael Dixon

I am a research associate at the Washington University Media and Machines Lab, where I work on various object detection and tracking applications under the direction of Dr. Robert Pless.

Selected Publications

An Efficient System for Vehicle Tracking in Multi-Camera Networks
Michael Dixon, Nathan Jacobs, and Robert Pless
ACM/IEEE International Conference on Distrubted Smart Cameras (ICDSC 2009)
(pdf)
Faster and More Accurate Face Detection on Mobile Robots Using Geometric Constraints
Michael Dixon, Frederick Heckel, Robert Pless, and William D. Smart
IEEE/RSJ International Conference on Robots and Systems (IROS 2007)
(pdf) (ppt)
Say Cheese!: Experiences with a Robot Photographer
Zachary Byers, Michael Dixon, William Smart, and Cindy Grimm
AI Magazine, 25(3): 37-46, 2004.
(pdf)
An Autonomous Robot Photographer
Zachary Byers, Michael Dixon, Kevin Goodier, Cindy Grimm, and William Smart
IEEE/RSJ International Conference on Robots and Systems (IROS 2003)
(pdf)

(View the complete list)

Projects

Vehicle Tracking in Multi-Camera Networks
The recent deployment of very large-scale camera networks has led to a unique version of the tracking problem whose goal is to detect and track every vehicle within a large urban area. To address this problem we have developed a tracking system that is able to exploit constraints inherent in urban environments (i.e. while there are often many vehicles, they follow relatively consistent paths) in order to detecting cars in a fixed scene and connect these detections into partial tracks. We have also derived extensions to a network flow based probabilistic data association model, which we use to connect these tracks between cameras.
For more details, see the following: (pdf) (slides) (slides+video)
Face Tracking
As part of a multidisciplinary effort, requiring diverse expertise in physiology and anatomy, hardware design, signal processing, and computer vision, we have helped to develop a system for the remote monitoring of human physiological data. To support this effort, we have developed software capable of automatically detecting and tracking human subjects and optimally targeting the sensor.
More details about this work can be found in the following presentations:
- Face alignment (slides) (slides+video)
- Automatic targeting (slides) (slides+video)
Aircraft Transparency Inspection
In collaboration with The Boeing Company, we developed the first fully automated system for evaluating optical distortions in aircraft windshields. To quantify the distortion caused by the windshield, the system compares high-resolution digital images of a large grid pattern viewed through the windshield to a reference image of the gridboard. To determine if the windshield is production-worthy, the resulting distortion map is then evaluated by an automated classifier, trained on more than 200 hand-labeled examples. The final system has been delivered and has been successfully integrated into Boeing's windshield production process.
A detailed description of the system is available in the following article, currently under review by the Journal of Machine Vision and Applications: (pdf)
The Robot Photographer
In this project, we developed a fully autonomous robot photographer, capable of taking candid, well-composed photographs of people at social events. The robot wanders around the event, constantly evaluating photographic opportunities. When it identifies a likely shot, it uses standard photographic rules (such as the rule of thirds) to compose the picture, moves into position, and takes the shot. The photographer was deployed successfully at a number of public events, including SIGGRAPH 2002, Washington University's 150th birthday celebration, a science writers reception, and an actual wedding.

Selected Publications

Projects

Vehicle Tracking in Multi-Camera Networks

Face Tracking

Aircraft Transparency Inspection

The Robot Photographer

Education