Detection of Subpixel Submerged Mine-Like Targets in WorldView-2 Multispectral Imagery
Abstract
The utility of satellite spectral imagery is analyzed for the detection of submerged mine like objects. The targets that were utilized were sub-meter in size and emplaced in the water off the coast of La Jolla, California for a barnacle study. There were three surface targets and three submerged targets that did not exceed three meters in depth. Two meter multispectral and half meter panchromatic WorldView-2 data were used for image processing and analysis. The multispectral data proved more useful than the higher spatial resolution panchromatic data (sub-pixel vs resolved).For the multispectral data, principal component analysis was the most successful of the techniques, locating the submerged and surface targets. Attempts to use the RX-UTD anomaly detector were less successful, because of excessive false positives. This study supports the concept that commercial remote sensing is a viable option to support mine countermeasures.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 2012
- Accession Number
- ADA567365
Entities
People
- Michelle R. Sandersfeld
Organizations
- Naval Postgraduate School