遥感影像大范围地表信息缺失区域的修复理论与方法研究

Remotely sensed images are often degraded by large areas of dead pixels or thick clouds, which greatly weaken the potential applications in various fields. Due to the limitations of the traditional technique frameworks, the existing inpainting methods for remote sensing images can not satisfy the practical requirements. This project is to research and develop robust inpainting methods for remote sensing images by adopting the latest theory in mathematics, statistics and optimization fields. Firstly, an inpainting method for remote sensing images without complementary information is developed using Grouplet transform which is able to accurately capture geometry flow of the images. Secondly, an inpainting method for remote sensing images with complementary spectrum information is developed using compressed sensing theory which has been validated to be excellent in high-dimensional signal processing. Thirdly, an inpainting method for remote sensing images with time complementary information is developed using non-local regularization constraint model. Lastly, an inpainting method with integration of temporal-spectral complementary information is developed by combining compressed sensing with non-local regularization model. This study can expand new theories and produce new methods for remote sensing image processing, significantly enhance the potential applications of remote sensing data, therefore have important theoretical and applied significance.

遥感影像经常受到大范围死像元、厚云的影响，大大限制了其在各领域的应用潜力。由于理论框架的局限性，现有的遥感影像修复方法远不能满足实际处理的需求。本项目拟引入数学、统计、最优化领域的最新理论，研究和发展遥感影像大范围地表信息缺失区域的修复理论与方法。充分利用Grouplet变换准确捕捉影像几何流的特点，发展无互补信息的遥感影像修复方法；充分利用压缩感知理论对非线性问题与高维信号处理的优势，建立基于谱段互补信息的影像修复方法；基于非局部正则化约束模型建立变分求解框架，发展基于时相互补信息的影像修复方法；结合压缩感知与非局部正则化修复模型，实现集成时-谱互补信息的影像修复。本项目研究和发展遥感影像信息处理的新理论与新方法，可以大幅提升遥感数据的应用潜力，具有重要的理论与应用意义。

遥感影像经常受到大范围死像元、厚云的影响，大大限制了其在各领域的应用潜力。由于理论框架的局限性，现有的遥感影像修复方法远不能满足实际处理的需求。本项目引入数学、统计、最优化领域的最新理论，从无互补信息的影像修复、基于时间互补信息的影像修复、基于光谱互补信息的影像修复以及集成时空谱互补信息的影像修复等方面开展了全面的研究，建立了一套较完备的遥感影像大范围地表缺失区域的修复理论和方法体系。研究成果成功用于MODIS、Landsat等遥感数据的死像元修复、影像厚云去除、以及应用于积雪产品、地表温度产品、臭氧产品、DEM产品等遥感参量产品的重建，能有效提升遥感数据和产品的应用潜力。. 按照既定的研究计划，项目圆满完成了课题设定的研究内容，实现了预期目标，并在研究的深度和广度上都比既定计划有了进一步的扩展。在项目资助下，项目组发表科研论文25篇，其中19篇SCI检索；撰写1部专著（即将出版）；申请/获批专利4项；发布开源软件1套；发布遥感产品1套；在人才培养方面，项目负责人入选国家优秀青年基金、万人计划“青年拔尖人才”等人才计划；项目组培养毕业博士7名、硕士1名，在读博士生2人、硕士生2人。