中文摘要：

针对湍流环境下红外成像的特点，提出了一种基于广义规整化的湍流退化图像盲复原方法。该方法在合理建立红外图像退化模型的基础上，根据极大似然估计准则来设计。为了抑制噪声同时有效地引入关于图像的先验信息，对传统的规整化方法进行了扩展，提出了广义规整化的策略，将规整化分为两个各有侧重点的层次，即复原过程中噪声的抑制和含噪条件下红外图像相关特征的保持，根据有限的先验知识将复原问题的求解转化为带规整化约束的最优化估计问题，进而复原湍流退化图像。

英文摘要：

A turbulence-degraded blind image restoration method based on generalized regnlarization was proposed. A appropriate turbulence-degraded model was established for infrared image and the entire restoration method was designed in the framework of maximum-likelihood estimation. To suppress noise and incorporate knowledge about infrared image more effectively, the proposed method generalizes original nonlinear regnlarization. The strategy of generalized regnlarization was suggested, which divides regnlarization into two parts with different emphasis on noise-suppression and image edge-preservation, and alternately estimates PSF and object image.