A novel method is proposed for azimuth sidelobes suppression using multi-pass squinted (MPS) synthetic aperture radar (SAR) data. For MPS SAR, the radar observes the scene with different squint angles and heights on each pass. The MPS SAR mode acquisition geometry is given first. Then, 2D signals are focused and the images are registered to the master image. Based on the new signal model, elevation processing and incoherent addition are introduced in detail, which are the main parts for azimuth sidelobes suppression. Moreover, parameter design criteria in incoherent addition are derived for the best performance. With the proposed parameter optimization step, the new method has a prominent azimuth sidelobes suppression effect with a slightly better azimuth resolution, as verified by experimental results on both simulated point targets and TerraSAR-X data.

Spaceborne global navigation satellite system reflectometry has been used to retrieve wind speed, but few studies have directly detected tropical cyclones from spaceborne GNSS-R observations. Moreover, it is challenging to solve the multivalued dependence of the spaceborne observable on wind speed to unambiguously and accurately retrieve wind speed in cyclone conditions. This paper presents a single-pass cyclone detector and coarse estimation approach of the tropical cyclone position using a spaceborne GNSS-R full delay-Doppler map (FDDM). When the specular point passes through a tropical cyclone, the FDDM asymmetry experiences an abnormal change. From the FDDM asymmetry sequence along the specular point trajectory, two subsequence features, including the slope and extremum difference, are defined, from which the tropical cyclone detector is proposed. The results from the simulation and the Cyclone GNSS (CYGNSS) data both indicate a good detection performance for tropical cyclones. The location corresponding to the peak detector is considered as a coarse estimation of the tropical cyclone position. The test result from the CYGNSS data shows that the mean error between the detected cyclone center and the International Best Track Archive for Climate Stewardship (IBTrACS) cyclone center is approximately 124.50 km. From the detector of tropical cyclones, a scene-classified wind speed retrieval model is proposed. The simulated and experimental results show that a better retrieval performance can be obtained at high wind speed (> 30 m/s) using the scene-classified model. Reductions of 10 m/s and 4 m/s in the root mean square errors (RMSEs) are obtained for the simulation and CYGNSS data, respectively. This work is meaningful for directly detecting tropical cyclones and retrieving high wind speed data using spaceborne GNSS-R in real time.

Synthetic aperture radar (SAR), with its all-weather and all-day active imaging capability, has become indispensable for geoscientific analysis and socio-economic applications. Despite advances in deep learning–based object detection, the rapid and accurate detection of small objects in SAR imagery remains a major challenge due to their extremely limited pixel representation, blurred boundaries in dense distributions, and the imbalance of positive–negative samples during training. Recently, vision–language models such as Contrastive Language-Image Pre-Training (CLIP) have attracted widespread research interest for their powerful cross-modal semantic modeling capabilities. Nevertheless, their potential to guide precise localization and detection of small objects in SAR imagery has not yet been fully exploited. To overcome these limitations, we propose the CLIP-Driven Adaptive Tiny Object Detection Diffusion Network (CDATOD-Diff). This framework introduces a CLIP image–text encoding-guided dynamic sampling strategy that leverages cross-modal semantic priors to alleviate the scarcity of effective positive samples. Furthermore, a generative diffusion-based module reformulates the sampling process through iterative denoising, enhancing contextual awareness. To address regression instability, we design a Balanced Corner–IoU (BC-IoU) loss, which decouples corner localization from scale variation and reduces sensitivity to minor positional errors, thereby stabilizing bounding box predictions. Extensive experiments conducted on multiple SAR and optical remote sensing datasets demonstrate that CDATOD-Diff achieves state-of-the-art performance, delivering significant improvements in detection robustness and localization accuracy under challenging small-object scenarios with complex backgrounds and dense distributions.

Deep learning (DL) has shown great capability in remote sensing and automatic target recognition (ATR). However, huge computational costs and power consumption are challenging the development of current DL methods. Optical neural networks have recently been proposed to provide a new mode to replace artificial neural networks. Here, we develop a residual optronic convolutional neural network (res-OPCNN) for synthetic aperture radar (SAR) recognition tasks. We implement almost all computational operations in optics and significantly decrease the network computational costs. Compared with digital DL methods, res-OPCNN offers ultra-fast speed, low computation complexity, and low power consumption. Experiments on the Moving and Stationary Target Acquisition and Recognition (MSTAR) dataset demonstrate the lightweight nature of the optronic method and its feasibility for SAR target recognition.