| 摘要: |
| 合成孔径雷达(SAR)三维成像技术具备高度向的分辨能力,能够有效解决二维SAR成像中的叠掩问题。然而,现有技术在数据获取方面存在周期较长、成本较高等问题,同时,重建结果易受噪声干扰,导致异常值的出现。针对这些挑战,本文提出了一种融合光学信息的阵列干涉SAR三维成像方法。该方法通过引入光学图像提供的先验几何信息,对SAR三维重建过程进行约束,从而拓展了数据源,降低了SAR数据获取成本,并有效地纠正了异常值。具体而言,该方法首先进行初步的三维成像,并利用初始的三维点云建立起光学图像与SAR图像之间的配准关系；接着,对光学图像进行几何结构分割,并将分割结果作为约束条件应用于SAR三维重建过程中,最终得到基于几何约束的三维重建结果。通过实测数据进行验证,结果表明该方法显著抑制了三维重建中的离群散射点,获得了同时包含光学纹理和SAR纹理的高质量三维点云,有望解决光学和SAR二维图像融合面临的异质性难题,为后续基于融合图像的目标解译奠定基础。 |
| 关键词: 三维成像 二维映射 图像配准 几何约束 |
| DOI： |
| 分类号:TN957.52 |
| 基金项目:国家自然科学基金项目（面上项目，重点项目，重大项目） |
|
| 3D Imaging of Array Interferometric SAR with Integrated Optical Information |
|
|
|
|
| Abstract: |
| Synthetic Aperture Radar (SAR) three-dimensional (3D) imaging technology offers vertical resolution and effectively addresses the issue of image stacking in SAR imaging.However, existing technologies face challenges such as long data acquisition periods, high costs, and the reconstruction results are prone to noise interference, which may lead to the presence of outliers. To overcome these limitations, this paper proposes a novel method that integrates optical information with array Interferometric SAR (array-InSAR) for 3D imaging. The proposed method incorporates prior geometric information from optical images to constrain the SAR 3D reconstruction process, thereby expanding the data sources, reducing the cost of SAR data acquisition, and effectively correcting outliers. Specifically, the method first performs an initial 3D imaging process and establishes the registration relationship between the optical and SAR images using the initial 3D point cloud. The optical image is then segmented based on its geometric structure, and the segmentation results are applied as constraints during the SAR reconstruction process. Finally, the 3D reconstruction results based on geometric constraints are obtained. Validation using real-world data demonstrates that the proposed method significantly suppresses outlier scatter points in the 3D reconstruction, resulting in high-quality 3D point clouds that simultaneously incorporate optical and SAR textures. This approach shows promise for addressing the heterogeneity challenges encountered in the fusion of optical and SAR 2D images, providing a solid foundation for subsequent target interpretation based on fused imagery. |
| Key words: 3D imaging 2D mapping image registration geometric constraints |