Offshore Platform Detection Based on Harris Detector and Intensity-Texture Image from Sentinel_2A Image

Efficiently locating offshore platforms is essential for management and safety maintenance of marine environment. Performance of the existing offshore platform detection methods with optical remote sensing image rely on large number of time-series images for false positive and false negative are serious in the step of capturing platform with single image. To improve the accuracy of platform detection with single image, a method based on Harris detector and intensity-texture feature image (HIT) is proposed. The intensity-texture feature image is first generated by combining intensity saliency feature and Gabor texture. Then, the Harris detector is employed to seek for platform candidates on the feature image. Eventually, offshore platforms were located according to the corner strength and geometrical morphology of the candidates. Experiments performed with Sentinel_2A images of different sea and weather condition demonstrated that the proposed method is more accurate with less false positive and false negative mistakes compared with the state-of-the-art global threshold method.