Segmentation of Multilook, Multifrequency, and Multipolarimetric SAR Data.
Abstract
This final report summarizes the findings of the research, "Segmentation of Multi-look, Multi-frequency and Multi-polarimetric SARt data." During the duration of the project, we have developed algorithms for (a) Markov Random Field based segmentation of high resolution SAR images, (b) detection of man-made features in SAR images and (c) labeling, as well as, grouping algorithms. These algorithms have been integrated to produce a 2-D site model of the given SAR image. The 2-D site model is an annotated description of the SAR image incorporating natural and man-made features such as trees, grass, water, open terrain, buildings, roads and shadows. Such site models are useful for delineating regions of interest (which serve as focus of attention mechanisms) and for providing local context in ATR algorithms.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 20, 1995
- Accession Number
- ADA303538
Entities
People
- Rama Chellappa
Organizations
- University of Maryland