基于关键帧间矩阵约束的多帧湍流图像盲复原  

Multi-frame Turbulence Image Blind Restoration Based on Matrix Constraint Between Key Frames

作　　者：鱼轮 李晖晖[2] YU Lun;LI Huihui(School of Electronic Information and Electrical Engineering,Shangluo University,Shangluo 726000;Laboratory of Information Fusion Technology and Ministry of Education,College of Automation,Northwestern Polytechnical University,Xi'an 710129)

机构地区：[1]商洛学院电子信息与电气工程学院,商洛726000 [2]西北工业大学自动化学院信息融合技术教育部重点实验室,西安710129

出　　处：《计算机与数字工程》2021年第6期1205-1210,共6页Computer & Digital Engineering

基　　金：国家自然科学基金项目(编号:61333017)资助。

摘　　要：针对传统多帧湍流图像复原耗时长及观测帧序列空间不齐时需匹配的问题,论文提出一种基于关键帧间矩阵约束的多帧湍流图像盲复原算法。该算法首先对观测序列自适应提取关键帧,然后在频域内分别构造估计原始图像和点扩散函数的代价函数方程,利用关键帧间矩阵约束指导点扩散函数复原,通过各向同性总变分模型约束指导图像复原,采用变量分离和增广拉格朗日方法对代价函数交替迭代优化,以达到快速性估计点扩散函数和恢复图像。仿真实验结果表明,论文提出的算法对图像复原质量有所提升,能求得更接近真实的点扩散函数值,且对未匹配的图像序列依然能达到良好的复原效果。Aiming at the problem that the traditional multi-frame turbulence image recovery takes a long time,and the obser⁃vation frame sequence space is not uniform,this paper proposes a multi-frame turbulence image blind restoration algorithm based on matrix constraint between key frames.Firstly,the algorithm adaptively extracts key frames from the observation sequence,and then the cost function equations for estimating the original image and the point spread function in the frequency domain are construct⁃ed,and the key inter-frame matrix constraint is used to guide the point spread function recovery,and the isotropic total variation model is passed to guide the image recovery,and the variable separation and augmented Lagrangian method are used to alternate it⁃erative optimization of the cost function to achieve fast estimation of the point spread function and recovery image.The simulation re⁃sults show that the proposed algorithm improves the image restoration quality,and it can obtain the point spread function value clos⁃er to the real one and still achieve a good recovery effect for the unmatched image sequence.

关 键 词：图像盲复原 关键帧间矩阵约束 交替迭代 点扩散函数 变量分离 

分 类 号：TN911.73[电子电信—通信与信息系统]