Space and Time Scale Characterization of Image Data in Varying Environmental Conditions for Better Scene Understanding
Abstract
Interpreting spatially and temporally changing scenes due varying environmental conditions and visual motion of objects within the field of view can pose serious challenges for rapid and robust scene understanding, particularly for problems of interest to the Army. Weather events, smoke, obscurants, or other changes in lighting and visibility can all affect image contrast and resolution, as can blurring caused by rapid movements or long exposure times in single frames and sequences of recorded images. To help mitigate some of the difficulties inherent in measuring and analyzing changing scenes, I propose that it is important to focus on the space and time scales of image data from the very beginning of the data collection process. This top-down approach not only helps to systematically characterize the measured data, but can help the end user determine which analysis or computer vision tasks are feasible with the available data. Alternately, this approach may be useful to predetermine what image resolutions are needed to enable more intelligent data collection. In this report, I begin to explore the space and time scale aspects of image data, discuss image motion characterization, and propose follow-on research studies to develop numerical algorithms and experiments to explore and analyze changing image scenes with new or existing data sets.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 2015
- Accession Number
- ADA622186
Entities
People
- Arnold D. Tunick
Organizations
- United States Army Research Laboratory