提出面向成像系统的光学和数字图像处理联合优化设计.根据泽尼克多项式模型分析了每种像差对成像清晰度的影响; 在变分贝叶斯框架下, 提出了基于加权双向差分先验模型的像差数字校正算法; 将像差分为易于数字校正和不易于数字校正两类, 建立基于像差选择性校正策略的光学-数字处理联合设计方法.采用联合设计方法设计了三镜片大像差光学系统, 其奈奎斯特频率处的MTF值约为0.50, 采用传统方法设计的六镜片光学系统MTF值约为0.53, 两系统成像质量相当, 本文方法可降低系统复杂度.

This paper is aiming at jointly optimizing the design and analysis of the optics and digital image processing for imaging systems. The influence of each aberration on the imaging sharpness is analyzed according to the Zernike polynomial model. A digital image processing algorithm is proposed for aberration correction, which is based on the variational bayesian framework and weighted bi-direction difference prior. Then the aberrations are divided into two categories: easy to digital correct and hard to digital correct, so an optics-digital processing co-design method can be established based aberration optional-correction. At last, a three-lens large aberration optical system is designed by the co-design method, its MTF value at the nyquist frequency is about 0.50; in contrast, a six-lens system is designed by traditional method, and its MTF value is about 0.53. These results show that the imaging quality of two system is similar, therefore, the proposed method would be capable of reducing the complexity of the optical system.