Since the 1980s, satellite-borne synthetic aperture radar (SAR) has been investigated for early warning and monitoring of marine oil spills to permit effective satellite surveillance in the marine environment.
Automated detection of oil spills from satellite SAR intensity imagery consists of three steps: 1) Detection of dark spots; 2) Extraction of features from the detected dark spots; and 3) Classification of the dark spots into oil spills and look-alikes. However, marine oil spill detection is a very difficult and challenging task. Open questions exist in each of the three stages.
In this thesis, the focus is on the first stage—dark spot detection. An efficient and effective dark spot detection method is critical and fundamental for developing an automated oil spill detection system. A novel method for this task is presented. The key to the method is utilizing the spatial density feature to enhance the separability of dark spots and the background. After an adaptive intensity thresholding, a spatial density thresholding is further used to differentiate dark spots from the background. The proposed method was applied to a evaluation dataset with 60 RADARSAT-1 ScanSAR Narrow Beam intensity images containing oil spill anomalies. The experimental results obtained from the test dataset demonstrate that the proposed method for dark spot detection is fast, robust and effective. Recommendations are given for future research to be conducted to ensure that this procedure goes beyond the prototype stage and becomes a practical application.
| Identifer | oai:union.ndltd.org:WATERLOO/oai:uwspace.uwaterloo.ca:10012/5009 |
| Date | 28 January 2010 |
| Creators | Shu, Yuanming |
| Source Sets | University of Waterloo Electronic Theses Repository |
| Language | English |
| Detected Language | English |
| Type | Thesis or Dissertation |
Page generated in 0.0019 seconds