Ship detection in optical remote sensing image based on visual saliency and AdaBoost classifier

In this paper, firstly, target candidate regions are extracted by combining maximum symmetric surround saliency detection algorithm with a cellular automata dynamic evolution model. Secondly, an eigenvector independent of the ship target size is constructed by combining the shape feature with ship histogram of oriented gradient (S-HOG) feature, and the target can be recognized by AdaBoost classifier. As demonstrated in our experiments, the proposed method with the detection accuracy of over 96% outperforms the state-of-the-art method. © 2017, Tianjin University of Technology and Springer-Verlag Berlin Heidelberg.