空间薄膜衍射光学成像系统图像退化特性及恢复方法

The diffractive membrane optical imaging system has been an important development trend for very large aperture and lightweight space camera. However, the aspects of physics-based imaging degradation characteristics and corresponding image restoration have not been investigated. In view of the problems, this study has the following main aspects: Firstly, imaging degradation rule and characteristics are explored based on the theory of diffractive membrane. On this basic, a new regularization model with multiple prior constraints is established based on sparse representation and regularization frame, according to multi-objective improvement including MTF and SNR improving, chromatic aberration correcting, edge and texture preserving as well as artifact suppressing. Moreover, the sparse representation approach of PSF is studied and then the corresponding sparse image prior is proposed aiming at the problems of the huge calculation induced by space-variant point spread function (PSF) image restoration for large aperture imaging system. And the solving method of the model is investigated aiming at the problems of multi-variable estimation, multi-norm coexistence and multi-regularizing parameter optimization. Then a new image restoration method based on multi-objective improvement is proposed. Finally, the performance of the proposed method is verified, which can provide scientific basis for practical application of and possesses potential application prospects on the parameters optimization of imaging system design. The work can fill the gap of exploratory foundation research in this field, and has the theoretical significance and engineering application value for the future space application of diffractive membrane imaging technology.

薄膜衍射光学成像系统已成为未来超大口径空间载荷的重要发展方向。而目前关于薄膜衍射系统图像退化特性及恢复方面的研究几乎空白。本项目以薄膜衍射成像理论为基础，开展薄膜衍射系统在轨成像质量退化机理及特性研究；针对退化特性，以同时提升图像传函和信噪比、校正色差、保持纹理细节及抑制伪像为目标，研究相应的正则化先验约束，同时针对大尺寸点扩散函数（PSF）空变图像恢复的巨大计算量问题，结合稀疏表示思想和有界变分理论，研究PSF的稀疏性先验，从而基于稀疏表示的正则化理论框架，建立融合多目标正则化先验约束的PSF空变图像恢复模型。针对模型中多变量估计及同时包含多种范数正则化项和参数的联合寻优问题，研究快速鲁棒的模型求解算法，提出图象恢复方法；最后开展实验验证，为方法应用及成像系统设计提供科学依据与理论指导。该项目研究将填补国内在该领域探索性基础研究的空白，对该项技术的未来应用具有理论研究意义和工程应用价值。

薄膜衍射光学成像系统已成为未来超大口径空间载荷的重要发展方向。而目前关于薄膜衍射系统图像退化特性及恢复方面的研究几乎空白。本项目从薄膜衍射光学成像系统在轨图像质量退化机理出发，分析得出了其具有衍射效率低、图像信噪比低、传函退化严重、PSF大尺寸空变等特性；针对退化特性，以同时提升图像传函和信噪比、校正色差、保持纹理细节及抑制伪像为目标，研究相应的图像先验和正则化约束项，同时针对大尺寸PSF空变图像恢复计算量巨大的问题，结合稀疏表示思想和有界变分理论，构建了PSF稀疏化表示先验，建立了融合多目标正则化先验约束的PSF空变图像恢复模型；并针对模型中多变量估计及同时包含多种范数正则化项和参数联合寻优的难题，研究了快速鲁棒的模型求解算法，从而提出了多目标多参量图像恢复方法。实验结果表明该方法能够提升图像传函和清晰度、解决PSF空变问题、保持图像细节的同时，有效抑制噪声和伪像，可有效支撑系统的地面图像处理应用。该项目研究将填补国内在该领域探索性基础理论研究的空白，对薄膜衍射光学成像技术的未来空间应用具有理论研究意义和工程应用价值。通过创新之处的总结凝练，发表SCI/EI收录论文7篇，申请发明专利7项。相关成果已应用于高分静止轨道光学卫星系统任务分析与效能仿真，并正在指导后续航天装备发展型号顶层设计与关键指标论证。