Ship detection in high-resolution synthetic aperture radar (SAR) images has attracted great attention. As a popular method, a constant false alarm rate (CFAR) detection algorithm is widely used. However, the detection performance of CFAR is easily affected by speckle noise. Moreover, the sliding window technique cannot effectively differentiate between clutter and target pixels and easily leads to a high computation load. In this paper, we propose a new superpixel-based non-window CFAR ship detection method for SAR images, which introduces superpixels to CFAR detection to resolve the aforementioned drawbacks. Firstly, our previously proposed fast density-based spatial clustering of applications with noise (DBSCAN) superpixel generation method is utilized to produce the superpixels for SAR images. With the assumption that SAR data obeys gamma distribution, the superpixel dissimilarity is defined. Then, superpixels can be accurately used to estimate the clutter parameters for the tested pixel, even in the multi-target situations, avoiding the drawbacks of the sliding window in the traditional CFAR. Moreover, a local superpixel contrast is proposed to optimize CFAR detection, which can eliminate numerous clutter false alarms, such as man-made urban areas and low bushes. Experimental results with real SAR images indicate that the proposed method can achieve ship detection with a higher speed and accuracy in comparison with other state-of-the-art methods.

Built-up areas (BAs) information acquisition is essential to urban planning and sustainable development in the Greater Bay Area in China. In this paper, a pseudo-Siamese dense convolutional network, namely PSDNet, is proposed to automatically extract BAs from the spaceborne synthetic aperture radar (SAR) data in the Greater Bay Area, which considers the spatial statistical features and speckle features in SAR images. The local indicators of spatial association, including Moran’s, Geary’s, and Getis’ together with the speckle divergence feature, are calculated for the SAR data, which can indicate the potential BAs. The amplitude SAR images and the corresponding features are then regarded as the inputs for PSDNet. In this framework, a pseudo-Siamese network can independently learn the BAs discrimination ability from the SAR original amplitude image and the features. The DenseNet is adopted as the backbone network of each channel, which can improve the efficiency while extracting the deep features of the BAs. Moreover, it also has the ability to extract the BAs with multi-scale sizes by using a multi-scale decoder. The Sentinel-1 (S1) SAR data for the Greater Bay Area in China are used for the experimental validation. Our method of BA extraction can achieve above 90% accuracy, which is similar to the current urban extraction product, demonstrating that our method can achieve BA mapping for spaceborne SAR data.