Invariant triangle-based stationary oil platform detection from multitemporal synthetic aperture radar data

摘要

An automatic algorithm for stationary oil platform detection from multitemporal synthetic aperture radar data is proposed. The proposed algorithm consists of the following two parts. (1) A two-parameter constant false-alarm rate (CFAR) algorithm is used to extract targets from the Environment Satellite (ENVISAT) advanced synthetic aperture radar (ASAR), in which the focus is to determine the appropriate parameters of CFAR, thus ensuring as few as possible false-alarm targets when sea-surface targets are effectively extracted. (2) A simple point cluster matching pattern is proposed based on an invariant triangle rule, by which targets extracted from multitemporal ENVISAT ASAR images are automatically matched for detection of stationary targets (e.g., oil platforms). This invariant triangle rule is that any three moving targets have an extremely low probability of maintaining a relative position in multitemporal images, whereas stationary targets can always maintain a fixed relative position. Even under high noise, this invariant triangle rule can be used to realize the target data matching with high robustness. The experiment shows that the false-alarm rate and the missing rate are relatively low when all the targets are detected. The proposed invariant-triangle-based point cluster matching pattern can conduct effective detection and monitoring of stationary oil platforms.