Recent News
Inaugural School of Engineering Teaching Innovation Fellows selected
February 2, 2024
UNM computer scientist wins NSF CAREER Award to optimize supercomputer performance
February 1, 2024
Hand and Machine Lab’s Experimental Clay Exhibition closing celebration Nov. 17
November 15, 2023
Moses selected as special assistant to the dean for educational initiatives
October 3, 2023
News Archives
Anomalous Change Detection in Remote Sensing Imagery
November 21, 2008
- Date: Friday, November 21st, 2008
- Time: 2 pm — 3:15 pm
- Place: ME 218
James Theiler
Space and Remote Sensing Sciences Group
Los Alamos National Laboratory
Abstract: The detection of actual changes in a pair of images is confounded by the inadvertent but pervasive differences that inevitably arise whenever two pictures are taken of the same scene, but at different times and under different conditions. These differences include effects due to illumination, calibration, misregistration, etc. If the actual changes of interest are assumed to be rare, then one can “learn” what the pervasive differences are, and can identify the deviations from this pattern as the anomalous changes.
While the straight anomaly detection problem is often expressed as a “one-class” problem, I will argue that it is really a two-class problem where the second (aka background or reference) class has implicit properties that are not always acknowledged. By adapting this background class, one can re-orient the machine learning methodology for anomaly detection to work for anomalous change detection.
I will describe some of these theoretical issues as well as more practical problems that arise in the anomalous change detection problem, and show some recent results from an ongoing project at Los Alamos.