In the domain of micromotion target recognition, target characteristics can be extracted through either broadband or narrowband radar echoes. However, due to technical limitations and cost constraints in acquiring broadband radar waveform data, recognition can often only be performed through narrowband radar waveforms. To fully utilize the information embedded within narrowband radar waveforms, it is necessary to conduct in-depth research on multi-dimensional features of micromotion targets, including radar cross-sections (RCSs), time frequency (TF) images, and cadence velocity diagrams (CVDs). To address the limitations of existing identification methodologies in achieving accurate recognition with narrowband echoes, this paper proposes a graph fusion network based on a cross-modal attention mechanism, termed GF-AM Net. The network first adopts convolutional neural networks (CNNs) to extract unimodal features from RCSs, TF images, and CVDs independently. Subsequently, a cross-modal attention mechanism integrates these extracted features into a graph structure, achieving multi-level interactions among unimodal, bimodal, and trimodal features. Finally, the fused features are input into a classification module to accomplish narrowband radar micromotion target identification. Experimental results demonstrate that the proposed methodology successfully captures potential correlations between modal features by incorporating cross-modal multi-level information interactions across different processing stages, exhibiting exceptional accuracy and robustness in narrowband radar micromotion target identification tasks.

Circular synthetic aperture radar (CSAR) has advantages of the omni-directional detection and high-resolution imaging. However, when CSAR detects the dense target in the observation scene, the phenomena such as the layover effect in the CSAR image may make mutual interference between the adjacent vehicles, which leads to the poor performance of dense target detection by traditional methods. To address this problem, this paper has proposed a high-precision imaging and dense vehicle target detection method for the airborne single-pass CSAR, which includes the CSAR high-precision imaging based on the extracted digital elevation model (DEM) information and high-precision dense vehicle target detection based on the CSAR image. For the high-precision imaging, the back-projection algorithm (BPA) is firstly used for the observation scene imaging, and then the DEM information of the vehicle target is extracted using the improved sub-aperture correlation method. Finally, the extracted DEM information is combined with the BPA for the high-precision imaging. For the high-precision detection, the obtained high-precision CSAR image is used to generate the maximally stable extremal region (MSER), then the extreme region is fused using the vehicle target length and width information, and finally the high-precision detection result of the dense vehicle target is obtained. Based on the public CSAR dataset from the American Air Force Research Laboratory, the experimental results show that the proposed method can obtain the high-precision CSAR images and improve the detection rate of dense vehicle target under the premise of ensuring a low false alarm rate.

The low-frequency ultra-wideband synthetic aperture radar (UWB SAR) has the ability of the foliage-penetrating and high-resolution imaging, which can detect the foliage-concealed target. However, due to low-frequency UWB SAR characteristics and forest detection environments, there are usually some non-target strong scattering points in the low-frequency UWB SAR images, which may increase the difficulty of the foliage-concealed target change detection. To solve the problem of the weak anti-jamming ability of the foliage-concealed target change detection, a foliage-concealed target change detection scheme based on the convolutional neural network (CNN) in the low-frequency UWB SAR images is proposed, which combines the traditional image difference method and deep learning method. Firstly, a relatively low inspection threshold is set for the target change detection based on the image difference method, which can obtain a lot of the position information of the detection point. Moreover, for the target characteristics in the foliage-concealed scenarios, the corresponding data extraction and enhancement techniques are used to effectively extract the detection point samples from the detection image and reference image, which can prevent the overfitting of the model training caused by the sample scarcity. Finally, the samples of the detection points are input to the target classification network with the double input and single output for the classification training and testing. Experimental results tested on the CARABAS-II SAR dataset demonstrate the correctness and effectiveness of the proposed scheme, which has the better change detection performance and anti-interference capability.

Multi-baseline circular synthetic aperture radar (MB CSAR) can be applied to obtain a three-dimensional (3D) image of the observation scene. However, the phase error caused by radar platform motion or atmospheric propagation delay restricts its 3D imaging capabilities. The phase error calibration of MB CSAR data is an essential step in the 3D imaging procedure due to the limited accuracy of positioning sensors. Phase gradient autofocus (PGA) is widely utilized to estimate the phase errors but is subject to shifts in the direction perpendicular to the line of sight and long iteration time in some sub-apertures. In this paper, an improved PGA method for MB CSAR 3D imaging is proposed, which can suppress the shifts and reduce computation time. This method is based on phase gradient estimation, but the prominent units are selected with an energy criterion. Then, weighted phase gradient estimation is presented to suppress the influence of prominent units with poor quality. Finally, a contrast criterion is adopted to reach faster convergence. The experimental results based on the measured MB CSAR data (Gotcha dataset) demonstrate the validity and feasibility of the proposed phase error calibration method for MB CSAR 3D imaging.