基于多分辨率盲目去卷积的气动光学效应退化图像复原算法

Investigation of Restoration Algorithm for Degraded Images Caused by Aero-optics Effects Using Multi-resolution Blind Deconvolution

为了有效地恢复气动光学效应退化图像 ,该文提出了一种基于两帧退化图像的多分辨率盲目去卷积复原算法 .该算法在大气流场的光学退化模型未知的情况下 ,采用两帧退化图像数据来估计点扩展函数值 .为了克服噪声的干扰 ,该文将点扩展函数非负性和光滑性约束转化为数学上可表达的惩罚项 ,融合到目标函数中 ,建立了一个基于各向异性调整的目标函数表达式 ,采用约束优化原理迭代估计点扩展函数值 .同时 ,针对复原图像的快速需求 ,将多分辨率技术用在点扩展函数的最优估计和图像复原过程中 ,提出了基于图像多分辨率的盲目去卷积复原新算法 .在微机上进行了一系列的图像恢复实验和对比分析 ,实验结果表明该文复原算法十分有效 ,在抑制噪声和提高复原速度方面取得了明显的效果 .

For the restoration of the degraded images caused by aero-optics effects of atmospheric turbulence, this paper presents an effective algorithm, multi-resolution blind deconvolution based on two frames of degraded images, in which the PSFs can be estimated by use of the two degraded images under the condition of the degraded models being unknown. In order to overcome the interference of noise, the constraints of non-negativity and smoothness of the PSFs are converted into the penalty terms expressed in mathematics and an objective function based on anisotropic regularization is built by adding the penalty terms into it. Thus, the PSFs' values can be estimated in an iteration manner by use of the constraint optimal principle, and then the images can be restored by the filter method. Meanwhile, for the requirement of fast restoration, an algorithm based on multi-resolution blind deconvolution is proposed by employing the image multi-resolution technology in the process of optimization estimation of PSFs and restoration. A series of restoration experiments and some comparison analysis have been performed in a microcomputer. The experiments results illustrate that the proposed algorithm is efficient with obvious effects on the aspects of resisting noise and speeding up the restoration.