Evaluation of the Navy's Semi-Automated Mesoscale Analysis System (SMAS)
Abstract
The feasibility of automated analysis of satellite imagery for positions of mesoscale ocean features has been demonstrated. Weaknesses in Version 1.0 have been identified and now form the rational for further research. Specifically, improvements have been initiated in the area of segmentation, where a new segmenter utilizing both edge and region information is under development. This first evaluation has also revealed that the feature labeling algorithm is too closely tied to the initial guess. Improvements in this area are also underway. A new approach to feature labeling using genetic algorithms is also under investigation.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 1990
- Accession Number
- ADA237978
Entities
People
- Ronald J. Holyer
- Sarah H. Peckinpaugh
Organizations
- United States Naval Research Laboratory