At the present time, most of existing security systems only detect and track targets in line-of-sight (LOS). However, in practice, the locations of targets are often out of the line of sight. This article focuses on the non-line-of-sight (NLOS) moving target detection with low-power transmission signals by reflection. There are two key problems, the weak target echo signal and the multipath effect. In terms of the issues, this paper constructs the echo signal model of the NLOS target. On the basis of the echo model, the detection method of NLOS moving target based on millimeter-wave radar comes up, which is of great theoretical value and important practical significance for indoor security. This paper innovatively applies the polynomial fitting method to suppress static noise and range gating method to suppress noise from other range gates. Then, the location and velocity of the target are estimated by two-dimensional fast Fourier transform (FFT) and the multiple signal classification (MUSIC) method. Furthermore, in order to verify the accuracy of the NLOS target echo signal model proposed in this paper, we respectively simulated two important parts of the signal in the model, the target echo signal and the direct backscattered signal of the intermediate interface, both of which are multipath signals. We counted the echo path length distribution in these two parts, and applied the NLOS target detection method to process them respectively. In addition, we also simulated the NLOS target echo signal and obtained actual data in the actual scene, and processed both the simulated data and the actual data. Comparing the results of target detection with and without denoising methods, the effectiveness of the two denoising methods proposed in this paper is verified.
23 October 2025 in IEEE Transactions on Geoscience and Remote Sensing
In Synthetic Aperture Radar (SAR) automatic target recognition (ATR) tasks, due to the SAR imaging mechanism and the characteristics of target scattering structures, the image features of a target under a single observation angle may be insufficiently expressed, and the image features under different observation angles may vary significantly, which makes target recognition using single-view SAR images challenging. Therefore, acquiring multi-view images of the target for joint recognition has become a promising method. However, existing multi-view recognition models suffer from problems such as insufficient learning of multi-view features, inapplicability under incomplete multi-view conditions, and scarcity of multi-view training data, restricting the performance and practicality. In this paper, we establish a new multi-view SAR target recognition framework to uniformly address the above issues. First, we construct a semantic-view dual-encoder feature embedding fusion network (SeVi). By performing classification modeling on the observation views of images to introduce view prior information, the model is guided to acquire and fuse view-aware features. Second, we propose a prototype-based missing view completion strategy, enabling the model to adapt to complex incomplete multi-view scenarios while maintaining high performance. Third, we put forward a non-strict multi-view recognition data construction method based on semi-supervised classification, which efficiently acquires data to support the training of multi-view recognition models. Experiments are conducted on multiple vehicle and aircraft target recognition datasets, including ATRNet-STAR, MSTAR, and SARAircraft. The results demonstrate that the proposed model can effectively capture and fuse view-aware features, achieving leading performance compared with existing SAR multi-view target recognition methods. Moreover, it maintains high performance even under complex incomplete multi-view complex conditions, which verifies the robustness and flexibility of the proposed multi-view recognition framework.
28 June 2023 in IEEE Transactions on Geoscience and Remote Sensing
The ground-based arc synthetic aperture radar (ArcSAR) can realize panoramic observation by utilizing the circular motion of the antenna. Since the circular trajectory of the antenna brings complicated high-order range-azimuth coupling in echo signal, far-field approximation, reference range approximation, or second-order series approximation to the range model are commonly applied for the conventional fast imaging algorithms to decouple the range-azimuth coupling. However, when a wide-beam millimeter wave radar system is adopted to realize centimeter-level high-resolution imaging for near-field observation, these approximate treatments will cause severe defocusing in the image. In this paper, an improved range-Doppler (RD) imaging algorithm is proposed. The high-order Taylor series approximation to the range model is adopted to meet the error requirement in the near-field wide-beam condition. By using the series inversion method, the analytical expressions for range migration and azimuth matched filter in range-Doppler domain are derived to achieve range-azimuth decoupling and precise azimuthal focusing within the whole range swath. Simulations and real-data experiments demonstrate that the proposed algorithm can achieve fast and accurate imaging which can greatly support the millimeter wave ArcSAR in near-field observation applications.
27 January 2025 in IEEE Transactions on Geoscience and Remote Sensing
Scene matching which involves establishing correspondences between multiple images of the same location, poses significant challenges for Synthetic Aperture Radar (SAR) images due to the anisotropic scattering prosperities of SAR targets; variations in looking and azimuth angles further complicate the matching process. To address these issues, a matching algorithm is proposed based on a coarse-to-fine framework. First, a coarse matching employing Normalized Cross Correlation (NCC) with sliding window is applied to filter out irrelevant regions, reducing distractions and shortening the processing time. Subsequently, a Siamese neural network, incorporating ResNet-50 and Convolutional Block Attention Module (CBAM) for enhanced feature extraction, is introduced to learn and discern differences between inputs. The effectiveness and robustness of the proposed method are validated through extensive experiments using a self-made dataset derived from Umbra Satellite.