Feature Extraction for Oceanographic Application

Abstract

Oil spill is the release of petroleum especially in water due to human activities or plant and animal growth. Oil spill is thus a form of pollution. It mainly occurs due to release of fuel from large ships such as bunker fuel. Cleanup and recovery from oil spills depend on factors such as type of oil, temperature of the water and the type of shoreline. The basic oil spill detection methods are: 1. In situ techniques (portable detection techniques) in which Researchers need to travel to the place with sensors and collect the data. Autonomous underwater vehicles (AUV) are used for the traveling purpose. 2. Remote sensing techniques (satellite sensing) in which the sensors are mounted on satellites to detect oil spills. In this thesis oil spill detection with remote sensing technique is explained. SAR images are First segmented using Hysteresis thresholding methods and then some of the features are extracted. After that the classification is done using decision tree algorithm. ImageJ which is a very powerful, open source, and java based tool is customized for oil spill detection process.