Oil tank detection via target-driven learning saliency model

Oil tanks detection is still a challenging task due to the complicated background in high-resolution remote sensing images. In this paper, we propose a new oil tanks detection approach based on target-driven learning saliency model (TDL). This model introduces the target-driven circular feature map to saliency model taking the value of circular density as a new weighting term of region contrast. Then we obtain the initial saliency map by optimizing the region contrast. For extracting salient target regions accurately, a strong classifier constructed by boosting algorithm, is used to obtain the global saliency map. Especially, all the training samples are determined by the initial saliency map. Then the two ways saliency maps are integrated to improve the detection performance. Extensive experiments are performed on the dataset containing 270 images of oil tanks differing in size, luminance and viewpoint, and the results show that the new method is effective in detecting oil tanks. Moreover, quantitative analyses verify that the method outperforms six state-of-Art saliency models and one oil tanks detection method.