Share Email Print
cover

Proceedings Paper

Automated oil spill detection with multispectral imagery
Author(s): Brian N. Bradford; Pedro J. Sanchez-Reyes
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

In this publication we present an automated detection method for ocean surface oil, like that which existed in the Gulf of Mexico as a result of the April 20, 2010 Deepwater Horizon drilling rig explosion. Regions of surface oil in airborne imagery are isolated using red, green, and blue bands from multispectral data sets. The oil shape isolation procedure involves a series of image processing functions to draw out the visual phenomenological features of the surface oil. These functions include selective color band combinations, contrast enhancement and histogram warping. An image segmentation process then separates out contiguous regions of oil to provide a raster mask to an analyst. We automate the detection algorithm to allow large volumes of data to be processed in a short time period, which can provide timely oil coverage statistics to response crews. Geo-referenced and mosaicked data sets enable the largest identified oil regions to be mapped to exact geographic coordinates. In our simulation, multispectral imagery came from multiple sources including first-hand data collected from the Gulf. Results of the simulation show the oil spill coverage area as a raster mask, along with histogram statistics of the oil pixels. A rough square footage estimate of the coverage is reported if the image ground sample distance is available.

Paper Details

Date Published: 5 May 2011
PDF: 11 pages
Proc. SPIE 8030, Ocean Sensing and Monitoring III, 80300L (5 May 2011); doi: 10.1117/12.883393
Show Author Affiliations
Brian N. Bradford, ITT Corp. Geospatial Systems (United States)
Pedro J. Sanchez-Reyes, ITT Corp. Geospatial Systems (United States)


Published in SPIE Proceedings Vol. 8030:
Ocean Sensing and Monitoring III
Weilin W. Hou; Robert Arnone, Editor(s)

© SPIE. Terms of Use
Back to Top