This article proposes a generalized compact polarimetry (GCP) mode along with two optimal polarization mode selection parameters to address the challenges of polarization mode selection in classification tasks across diverse scenarios. Theoretically, we conduct an in-depth analysis of the differences between circular and linear transmit polarizations, demonstrating their fundamental equivalence in terms of information content. For the first time, we propose that different classification tasks require different optimal polarization modes, and the optimal transmit polarization mode may lie in the elliptic polarization domain of synthetic aperture radar (SAR) systems rather than traditional circular compact polarimetric (CP) or linear dual-polarization (DP) modes. The proposed approach is validated using full-polarimetric SAR data from San Francisco and Hainan, showing that the optimal elliptical polarization mode achieves classification accuracies that are 2% to 42% higher than those of traditional CP or DP modes for certain categories, and performs comparably to full polarization. This improvement in accuracy stems from the interaction between the transmit polarization and the target scene, rather than advancements in classification algorithms. Using the two proposed parameters, the overall and category-specific classification performance of GCP modes can be effectively evaluated, enabling the identification of the optimal polarization mode for a given task. These findings provide significant insight into the design of future polarimetric SAR systems and offer new perspectives and directions for mission planning and mode selection for on-orbit satellites.

Graph Attention Network (GAT) has demonstrated better performance in many graph tasks than previous Graph Neural Networks (GNN). However, it involves graph attention operations with extra computing complexity. While a large amount of existing literature has researched GNN acceleration, few have focused on the attention mechanism in GAT. The graph attention mechanism makes the computation flow different. Therefore, previous GNN accelerators can not support GAT well. Besides, GAT distinguishes the importance of neighbors and makes it possible to reduce the workload through runtime tailoring. We present NTGAT, a software-hardware co-design approach to accelerate GAT with runtime node tailoring. Our work comprises both a runtime node tailoring algorithm and an accelerator design. We propose a pipeline sorting method and a hardware unit to support node tailoring during inference. The experiments show that our algorithm can reduce up to 86% of aggregation workload while incurring slight accuracy loss (<0.4%). And the FPGA based accelerator can achieve up to 3.8× speedup and 4.98× energy efficiency comparing to the GPU baseline.

As a new algorithm of graph embedding, graph neural networks (GNNs) have been widely used in many fields. However, GNN computing has the characteristics of both sparse graph processing and dense neural network, which make it difficult to be deployed efficiently on the existing graph processing accelerators or neural network accelerators. Recently, some GNN accelerators have been proposed, but the following challenges have not been fully solved: 1) the minibatch GNN inference scenario has the potential of software and hardware co-design, which can bring 30% computation amount reduction, and this is not well utilized. Besides, the cost of message flow graph construction is large and may account for more than 50% of the total delay; 2) the feature aggregation has a large amount of data access and relatively small amount of computation, which leads to low on-chip data reuse, only 10% of dense computing; and 3) without the optimization of sparse computing units, simple memory bank and cross bar architecture can easily lead to bank access conflict and load imbalance, reducing the utilization of computing units to less than 60%. In order to solve the above problems, we propose a algorithm-hardware co-design scheme to accelerate GNN inference, which includes three technologies: 1) a reuse-aware sampling method is proposed for minibatch inference scenarios, which reduces 30% of the calculation and improves the on-chip reusability of local data; 2) through the nodewise parallelism-aware quantization, the features and weights are quantized to integers with eight or four bits, which reduces the amount of memory access by at least four times; and 3) an accelerator supporting the above technologies is designed and evaluated, and different operations are supported by the sampling-inference integration architecture. The multibank on-chip memory pool is designed to support data reuse, and edge stream reordering is used to reduce data access conflicts, improving the utilization of computing units by$1.5\times $. Combined with the above technologies, the experiments show that our design achieves$9.2\times $speedup and$29\times $energy efficiency improvement compared with the Deep Graph Library framework running on servers equipped with CPU and GPU.

The channel imbalance and crosstalks are the two major factors for the polarimetric calibration of the circular quad-polarimetric (CQP) synthetic aperture radar (SAR) systems. In existing methods using distributed targets, the latter is usually ignored, which may lead to unbearable errors in target classification, surface parameter inversion, and so on. To address this issue, this article proposes a modified iterative calibration method using distributed targets that satisfy quasi-azimuth symmetry to calibrate both the channel imbalance and crosstalks of the CQP SAR systems. First, a stable distributed target selection strategy is proposed based on the correlation coefficient of LL polarization and RR polarization, the cross- and co-polarization backscatter ratio, and the equivalent number of looks (ENL). These parameters are insensitive to polarization distortion, and their typical ranges are determined via numerical simulations. Their combination helps select the targets that satisfy the quasi-azimuth symmetry, which is critical for calibrating the crosstalks. Then, the calibration can be conducted using the selected target. Finally, the phase ambiguity of the receive channel imbalance ratio, commonly found in distributed-target-based algorithms, is eliminated using a dipole target. Through the calibration of Gaofen-3 data and the statistical analysis of residual distortion, the effectiveness of the proposed method is verified.

In this paper, we investigate the potential of polarimetric synthetic aperture radar (SAR) in identifying flat areas using fractal dimension and polarimetric scattering similarity. A two-step method is proposed, including rough selection and fine selection. First, rough selection is performed by calculating the fractal dimension of the radar backscattered total power image. Then for each candidate region, the fine selection is conducted using polarimetric scattering similarity parameters. Furthermore, the effectiveness of the method is verified by GF-3 quad-polarimetric SAR data and SRTM1 DEM data in desert areas of China. Results show that for the final selected flat area (320 × 320 m), the maximum elevation deviation is 3.39 m and the elevation standard deviation is 0.72 m. Therefore, without depending on additional DEM data, the proposed method can effectively achieve flat areas identification, which can be helpful for the future application of polarimetric SAR data in the Moon.

Due to the large equivalent transmit crosstalk in hybrid quadrature polarimetric SAR system, traditional calibration methods using azimuthally symmetric distributed targets (AS-Targets) cannot work. In this manuscript, a preprocessing method of AS-Targets' covariance matrix is proposed, which makes equivalent crosstalk after processed become small to achieve system calibration through the AS-Targets. Simulation and airborne SAR data testing verify the reliability of the proposed method. The results show that the method presented in this manuscript can well calibrate the hybrid quadrature polarimetric SAR system using AS-Targets.

Polarimetric synthetic aperture radar (SAR) has been extensively used in various remote sensing applications. In this article, a unified framework is designed to compare the classification performance of different polarimetric systems, which include quad-polarimetric (QP), compact-polarimetric (CP), and dual-polarimetric (DP). To avoid problems, such as the lack of uniform standards in feature extraction, the classification algorithm is directly based on the statistical characteristics of the coherency/covariance matrix and is implemented by extending the Wishart mixture model (WMM). The GF-3 data set in San Francisco and the AIRSAR agricultural data set in Flevoland are used in the experiment, and the following conclusions are generated. QP can achieve the highest classification accuracy in all classification tasks. When distinguishing three typical classes (water, urban, and vegetation) with very different scattering characteristics, the performance of different polarimetric systems is similar, and QP has only a slight advantage. For classification tasks of different classes with similar scattering characteristics, CP performs better in agricultural scenes, and the overall accuracy (OA) is only reduced by 3%–4% compared with QP. DP performs better in urban scenes, and OA is only reduced by 1%–3% compared with QP. These conclusions can provide guidance for future payloads’ design and the choice of polarimetric operation mode for existing multi-polarimetric SAR systems to achieve the purpose of giving full play to the advantages of different polarimetric systems.

This paper established a unified framework to compare the capabilities of different polarimetric SAR systems in target detection based on polarimetric covariance matrix. The framework contains two different strategies. The first is based on the original polarimetric matrix and the second is based on the main scattering matrix. An improved algorithm is proposed based on the second strategy. GF-3 quad-polarimetric data is used to compare the performance between different algorithms, the results show that the proposed algorithm can obtain the highest signal-to-clutter ratio (SCR). The performance of different polarimetric systems is further compared, and the results show that the quad-polarimetric (QP) performs best, compact-polarimetric (CP) performance is similar to QP, and dual-polarimetric (DP) performs worst.