Development of Crude Oil Spill Imaging Database for Effective Quick Spill Detection and Remediation Procedures

Abstract

Crude oil spills need to be detected as quickly as possible in order to prevent damage to the environment, aquatic and land animals, humans and financial losses. Visible sensors which are passive sensors operating in the visible region of the light, are still widely used in oil spill remote sensing despite many shortcomings, including the fact that there are no established methods to ensure the positive detection of an oil spill in visible sensor images. There is a need for a publicly available database of ground truth crude oil spill visible images for researchers to develop effective automated spill detection algorithms for images in the visible spectrum. This paper presents the design and development of the first Crude Oil Spill Imaging Database available to the Oil and Gas Industry to assist in equipping and training them in swift and automated crude oil spill detection, estimation, classification, and recommendation of the appropriate clean-up and remediation procedure for each type of spill encountered, ensuring full remediation of our environment and protecting our health. The database currently contains 104 images of simulated ground truth visible crude oil spill images. Future work includes increasing the number of visible crude oil spill images to 500, and the inclusion of spill images taken with infrared sensors, optical, laser fluourosensors, synthetic aperture radar (SAR) sensors and thermal-infrared sensors.