FINet: A Feature Interaction Network for SAR Ship Object-Level and Pixel-Level Detection

Deep learning-based detection methods have achieved great success in ship target detection in synthetic aperture radar (SAR) images. However, due to the interference of imaging mechanism, speckle noise, and sea and land clutter, ship detection in SAR images still suffers from difficult interpretation. It is found that most ship detection algorithms focus on object-level detection while ignoring pixel-level information. In order to further improve the recognition effectiveness and positioning accuracy of ships in SAR images, we present a novel ship detection method based on a feature interaction network (FINet) in SAR images from the perspective of object-level and pixel-level. FINet consists of an object-level detection network and a pixel-level detection network. The information of the two branches is fused through the feature interaction module (FIM), and then, the object-level information and pixel-level information are enhanced by the feature guidance module (FGM). Finally, FINet utilizes object-level and pixel-level detection heads for prediction and regression to obtain object-level classification accuracy, positioning bounding box coordinates, and pixel-level binary classification results. The experimental results demonstrate that the classification effectiveness and localization accuracy of FINet are better than those of the comparison algorithms, and FINet achieves the best performance.

I. Introduction

Synthetic aperture radar (SAR), as an active microwave detection sensor, can provide shareable geospatial intelligence at any time and under any weather conditions, so it plays an important role in military and civil ship detection [1], [2], [3], [4], [5], [6]. In maritime management, SAR can monitor and manage ships entering and leaving ports throughout the day, helping to improve transportation efficiency and reduce maritime traffic accidents; in military affairs, SAR can detect and warn enemy ships, enhance situational awareness, and gain tactical advantages. However, due to the influence of imaging mechanisms, sea and land clutter, and target characteristics, ship detection still has problems in classification effectiveness and positioning accuracy. From the perspective of electromagnetic scattering mechanism, the sea scattering is mainly surface scattering, while the ship target has a different scattering mechanism from the sea surface. The scattering intensity of ship is large, and the scattering intensity of sea surface is small, so it is easy to distinguish ship target from sea clutter. The scattering of islands and land is strong, which has a great impact on ship detection. The traditional ship detection method needs to do sea–land segmentation preprocessing to eliminate the land interference. For deep learning, the features of ships and islands and land are different. Due to the complex structure, the ship target can be considered as a collection of different “angle scatterers.” Through the ground truth of specific mark, the network can learn the feature of the ship, so as to eliminate the interference of islands and land. Therefore, ship detection based on deep learning in SAR images is still attracting the attention of researchers.