Ship Detection in SAR Using Extreme Learning Machine

Abstract

Ship detection is an important issue in many aspects, vessel traffic services, fishery management and rescue. Synthetic aperture radar (SAR) can produce real high resolution images with relatively small aperture in sea surfaces. A novel method employing extreme learning machine is proposed to detect ship in SAR. After the image preprocessing, some features including entropy, contrast, energy, correlation and inverse difference moment are selected as features for ship detection. The experimental results demonstrate that the proposed ship detection method based on extreme learning machine is more efficient than other learning-based methods with prior performance of accuracy, time consumed and ROC.