In recent years, it has been a research hotspot to apply big data-driven deep learning methods to Synthetic Aperture Radar (SAR) target recognition with limited data. However, the problem caused by the long-tailed characteristics of SAR data has long been ignored. Specifically, a majority of data samples are concentrated in a few categories, leading to a skewed distribution of data. This skewed distribution can cause learning bias towards the majority class, which can subsequently degrade the recognition performance of the minority class. This issue is further exacerbated in limited sample conditions for SAR target recognition. After conducting research on target recognition for long-tailed natural images, this study has found that the existing methods used in this field cannot be easily applied to SAR target recognition. The primary reason is that SAR image data exhibit simultaneous and complex inter-class and intra-class long-tailed distributions. In response to this issue, we proposes the use of a multi-branch expert network and dual-environment sampling to address the long-tail problems in both inter-class and intra-class scenarios. The proposed method outperforms popular long-tailed target recognition methods on the long-tailed versions of the MSTAR and FUSAR datasets.
18 September 2020 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Image registration is an important processing step in synthetic aperture radar (SAR) image applications, such as change detection and elevation extraction. The cross-correlation method is widely employed to find the matching points to realize image registration due to its effectiveness and simplicity. However, the large number of pixel operations and whole image sliding operations make it a computationally intensive problem, and it is difficult to adapt to the situation of increasing amount and volume of SAR images. GPU-based high-performance computing methods are usually used because of their high parallelism and efficiency. However, most of these methods do not maximally optimize the computing process according to the characteristics of GPU architecture nor do they reduce the calculation amount of the registration process. In this paper, a swarm-intelligent GPU parallel pixel-level registration is proposed, which takes into account not only the acceleration of the correlation operation but also the reduction of searching times. First, for each correlation operation, the GPU parallelization is systematically optimized, including parallel reduction, bank conflict prevention, and instruction optimization. Second, the particle swarm optimization (PSO) algorithm is implemented by GPU to efficiently search the matching points based on the cross-correlation coefficients. In the process of calculation, the CPU and GPU have zero-copy, which realizes the complete parallelization of the registration. The experimental results show that the method can achieve 40 speedup for a product-level SAR image.
10 September 2024 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Filtering is a crucial step in SAR interferometric phase processing. In the context of rapid development in deep learning, the interferometric phase filtering networks have demonstrated more powerful filtering capabilities than spatial and transform domain filtering models. Inspired by the deep learning-based and non-local filtering models, this paper proposes a non-local SAR interferometric phase filtering model based on a partial Siamese network. Within this model, the interferogram is decomposed into multiple patches, and each patch undergoes filtering using a set of structurally identical encoder-decoder networks. In order to make better use of the non-local characteristics of the interferogram to improve the filtering effect, the encoder-decoder networks corresponding to different patches share a portion of their weights. Finally, the filtered patches outputted from each decoder are concatenated through an aggregation block to obtain the complete filtered interferogram. The advantage of this framework lies in its ability to fully leverage the powerful feature learning capabilities of neural networks while maximizing the utilization of non-local characteristics in the interferogram. The filtering performance of the proposed method is tested through simulated interferograms and real-world interferograms obtained from the Sentinel-1 and Gaofen-3 mission. The experimental results demonstrate that compared to traditional spatial and transform domain filtering methods, as well as state-of-the-art deep learning-based and nonlocal filtering methods, the filtering approach proposed in this paper achieves superior filtering performance.
In this paper, the study on DEM reconstruction using Gaofen-3 satellite repeat-pass SAR interferometry is presented. The coherence of the SLC images and residuals of interferogram provide a powerful guarantee for interferometric processing for Gaofen-3. Baseline estimation is the key factor to generate accurate DEM with SAR interferometry. This research uses the coarse orbit data and precise orbit data of Gaofen-3 to carry on the baseline estimation and uses the prior coarse DEM information to refine the baseline estimation. Then the topographic height is reconstructed with more precise spatial baseline. An experimental is conducted to verify the conclusion and some mountain and flatten region are chosen as the study area. The results show that, the DEM errors based on prior DEM method are rapidly converged compared with the ones based on the orbit data method. So benefit from the precise orbit data of the Gaofen-3 satellite, the accuracy of the baseline estimation can be improved further by using method based on prior DEM information, thus higher accuracy of DEM can be achieved. So that Gaofen-3 repeat-pass data have some good characteristics, such as high imaging quality, stable spatial baseline, high correlation, which has a good repeat-pass SAR interferometry application prospect.
20 March 2010 in Journal of Electronics & Information Technology