基于大气物理特性的光学遥感图像去雾化重建技术

The optical imaging technology takes one of the most important roles in the remote sensing of high resolution earth observation system, which is one of the nation's momentous scientific projects. Since the optical emission, dispersion and absorption of atmosphere have inevitable disturbance on the optical imaging spectral band, the optical remote sensing cameras are suffered greatly form atmospherically blurred when acquiring pictures on unfavorable weather condition, illumination condition, or at big side-view angle and large field of view imaging modes. The disturbance from atmosphere is becoming one of the important bottlenecks for acquiring high quality remote sensing images with high space, spectrum, radiance and time resolution. It is also one of the main hindrances for achieving images with poor illumination, big side-view and large field of view. .This research project is proposed to study the restoration technology of atmospherically blurred optical remote sensing images based on atmosphere physics models. The radiation, dispersion and absorption mechanism of aerosol and other small particles in typical weather and illumination condition will be studied. The spectral and radiant models of light rays which reach the image sensor after passing through atmosphere will be established. The image restoration methods for atmospherically blurred images will be studied based on the key physical models of the degenerated image caused by atmosphere disturbance. The ultimate target of this research project is to find the best ways for atmosphere correction of optical remote sensing systems. The achievements of this research works will have momentous significance for nation's optical remote sensing systems to improve the cameras' imaging abilities under unfavorable weather and illumination conditions, and to improve the earth observation image quality at big side-view angle and large field of view imaging modes.

光学遥感是高分辨率对地观测系统重大专项中信息获取的最主要技术手段之一。光学遥感成像过程中不可避免地受到大气辐射、散射、吸收等物理因素的干扰，特别是在非理想天气及光照条件和大侧摆、大视场成像模式下的成像质量受到很大影响。大气干扰成为制约遥感系统获取高分辨率（时间、空间、辐射、光谱）图像的重要因素之一，也成为遥感系统向高机动、宽覆盖方向发展的主要瓶颈之一。这方面传统的研究工作多基于对图像本身的锐化和增强处理。.本项目将通过对典型成像光照及气候条件下大气粒子、气溶胶的辐射、散射、吸收等机理研究，构建成像光线经过大气传输后到达探测器表面的辐射和光谱模型；基于大气物理特性对成像质量影响的关键模型，研究并提出光学遥感图像的去雾化重建方法，探求实现光学遥感图像大气校正的最佳技术途径。该项目的实施对提高我国光学遥感成像系统对天气和光照条件的适应能力，提高在大斜视、大视场等成像模式中的图像质量具有重要意义。

光学遥感成像光路中的大气辐射、散射、吸收等物理因素是制约其获取高分辨图像的重要因素之一。本项目基于大气物理特性，开展了针对典型成像光照及气候条件下大气粒子、气溶胶的辐射、散射、吸收等机理研究，构建了成像光线经过大气传输后达到光学遥感探测器表面的光谱和辐射模型，并基于其中的关键物理模型，提出了光学遥感图像的去雾化重建方法，为实现了光学遥感成像大气校正与图像质量提升，探索了一种有效的技术途径。.针对不同的应用场景和需求，本项目提出了一系列优化恢复算法，包括包含非均匀云雾的退化图像恢复算法，基于Angstrom指数的多光谱遥感图像去雾算法，基于MeanShift图像分割的去雾快速算法和基于遥感图像和地面观测点的大气分布估计算法。同时，基于实际场景数据库和真实地理大气信息，分析了同大气退化相关的遥感图像性质，结合神经网络学习框架，提出了遥感图像大气退化程度的量化评价方法。.本项目研究中发表SCI/EI收录论文15篇，申请发明专利5项，注册软件1项，培养博士研究生3名。.本项目部分研究成果已开始运用于我国高分辨率对地观测重大专项的光学遥感图像处理之中，对提高我国光学遥感成像系统对天气和光照条件的适应能力，提高在大斜视、大视场等成像模式中的遥感图像质量发挥了一定的作用