基于帧选择和极大似然估计的自适应光学图像多帧联合去卷积算法

吉林大学学报(理学版)

基于帧选择和极大似然估计的自适应光学图像多帧联合去卷积算法

张丽娟^1,2, 李东明^2,3, 杨进华², 邱欢¹, 刘颖⁴, 刘欢^3

1. 长春工业大学计算机科学与工程学院, 长春 130012; 2. 长春理工大学光电工程学院, 长春 130022;3. 吉林农业大学信息技术学院, 长春 130118; 4. 吉林财经大学管理科学与信息工程学院, 长春 130117

收稿日期:2016-07-22 出版日期:2017-09-26 发布日期:2017-09-26
通讯作者: 李东明 E-mail:ldmnuc@sina.com

Multi-frame Joint Deconvolution Algorithm for Adaptive Optics ImagesBased on FrameSelection and Maximum Likelihood Estimation

ZHANG Lijuan^1,2, LI Dongming^2,3, YANG Jinhua², QIU Huan¹, LIU Ying⁴, LIU Huan³

1. College of Computer Science and Engineering, Changchun University of Technology, Changchun 130012, China;2. School of OptoElectronic Engineering, Changchun University of Science and Technology, Changchun 130022, China;3. School of Information Technology, Jilin Agricultural University, Changchun 130118, China; 4. School of Management Science and Information Engineering, Jilin University of Finance and Economics, Changchun 130117, China

Received:2016-07-22 Online:2017-09-26 Published:2017-09-26
Contact: LI Dongming E-mail:ldmnuc@sina.com

摘要/Abstract

摘要： 为提高自适应光学(AO)图像的空间分辨率, 提出一种基于帧选择和极大似然估计的AO图像多帧联合去卷积算法. 该算法基于极大似然估计, 根据图像的高斯噪声模型建立多帧AO图像的联合对数似然函数. 首先对观测的多帧AO图像进行帧选择, 遴选高质量的降质图像; 然后结合观测条件和AO系统特性, 推导点扩散函数估计模型; 最后建立迭代求解公式, 得到多帧AO图像联合去卷积方法. 实验结果表明, 与基于期望值最大化的Richardson-Lucy算法(Richardson-Lucy EM算法)和基于合并惩罚函数的自适应应图像复原算法(CPF-Adaptive算法)相比, 该算法的峰值信噪比分别提高9%和5%, Laplace梯度模分别提高11%和8%, 且得到了较清晰的目标图像.

关键词: 点扩散函数, 自适应光学(AO), 极大似然估计, 图像复原, 帧选择技术

Abstract: In order to improve the spatial resolution of adaptive optics (AO) images, we proposed multi\|frame joint deconvolution algorithm for AO images based on frameselection and maximum likelihood estimation. The algorithm took maximum likelihood estimation as the basic principle. According to the Gaussian noise model of the image, the joint log likelihood function of multi\|frame AO images was established. Firstly, we selected frames of the observed multi\|AO
images and selected the high quality degraded images. Secondly, combined with the observed conditions and characteristics of AO system, the estimation model of point spread function was derived. Finally, we established iterative solution formulas, and obtained multi\|frame AO images joint deconvolution
method. The experimental results show that, compared with the RichardsonLucy EM or CPFAdaptive algorithm, the PSNR values of the proposed algorithm are increased by 9% and 5% respectively, and the Laplacian gradient moduli are increased by 11% and 8% respectively, and clearer target images are obtained.

Key words: frameselection technology, image restoration, point spread function (PSF), maximum likelihood estimation, adaptive optics (AO)

中图分类号:

TP751

张丽娟, 李东明, 杨进华, 邱欢, 刘颖, 刘欢. 基于帧选择和极大似然估计的自适应光学图像多帧联合去卷积算法[J]. 吉林大学学报(理学版), 2017, 55(05): 1199-1206.

ZHANG Lijuan, LI Dongming, YANG Jinhua, QIU Huan, LIU Ying, LIU Huan. Multi-frame Joint Deconvolution Algorithm for Adaptive Optics ImagesBased on FrameSelection and Maximum Likelihood Estimation[J]. Journal of Jilin University Science Edition, 2017, 55(05): 1199-1206.

[1]	冯海林, 郭甜. 基于Signature和Ⅱ型双删失系统寿命数据的统计推断[J]. 吉林大学学报(理学版), 2018, 56(5): 1125-1135.
[2]	赵波, 王纯杰, 李群, 佟知真. 左截断右删失数据下尺度参数与协变量相关时广义指数分布的估计及应用[J]. 吉林大学学报(理学版), 2018, 56(4): 864-870.
[3]	田家卓, 王德辉, 李雁林. (a,b)类分布的统计性质[J]. 吉林大学学报(理学版), 2016, 54(03): 480-486.
[4]	柯贤斌, 刘红卫, 游海龙. 基于梯度法的高效全局优化算法[J]. 吉林大学学报(理学版), 2016, 54(01): 40-44.
[5]	钟丽鸿, 胡成全, 金京姬. 基于RSSI极大似然估计定位算法的分析与实现[J]. 吉林大学学报(理学版), 2014, 52(03): 556-560.
[6]	施三支, 闫丽, 王泽升. p维TVPAR模型中参数的极大似然估计[J]. 吉林大学学报(理学版), 2013, 51(06): 1098-1100.
[7]	韩七星, 孙伟志. 随机食物有限种群模型参数估计的相合性及其渐近分布[J]. J4, 2012, 50(03): 482-.
[8]	赵志文,, 王思洋, 王瑞庭, 李玲. 定时截尾下具有部分缺失数据两个指数总体参数的估计与检验[J]. J4, 2009, 47(01): 26-30.
[9]	吴丹阳, 张忠波, 马驷良. 基于脊波变换的掌纹识别方法[J]. J4, 2008, 46(05): 944-946.
[10]	吴丹阳, 张忠波, 马驷良. 非接触式掌纹识别中的图像复原方法[J]. J4, 2008, 46(02): 293-295.
[11]	王宏志, 赵爽, 胡艳君. 基于L-曲线正则化的MAP超分辨率图像复原[J]. J4, 2008, 46(02): 275-278.
[12]	赵世舜, 宋洋, 宋立新. 对称熵损失下两个指数总体均值的序约束估计[J]. J4, 2007, 45(01): 44-48.
[13]	刘银萍. 具部分缺失数据两个指数总体的估计和检验[J]. J4, 2002, 40(03): 255-257.