AIRBASE DETECTION AND AIRSHIP RECOGNITION IN HIGH SPATIAL RESOLUTION REMOTE SENSING IMAGES

Automatic target detection in satellite images is a challenging problem due to the varying size, orientation and background of the target object. Airship recognition in remote sensing images is a meaningful task. It remains challenging due to the difficulty of obtaining appropriate representation of airships for recognition. To solve this problem, we propose a novel approach that is based on spatial frequency visual saliency analysis and convolutional neural networks. First, airbase is identified using spatial-frequency visual saliency analysis algorithm that is based on a CIE Lab color space to reduce the interference of backgroundand efficiently detect well-defined airbase regions in broad-area remote-sensing images. Second,an airship segmentation network is designed to obtain refined airship segmentation results. Then, a keypoints’ detection network is proposed to acquire airship’ directions and bounding boxes. At last, apply a template matching method to identify airships. Experiments show that the proposed method outperforms the state-of-the-art methods and can achieve more than 98% accuracy on the challenging data set.