顾及模糊核连通性的无人机图像半盲复原方法  

A Semi-blind Restoration Method of UAV Image Considering the Blurred Kernel Connectivity

作　　者：李佳田[1] 吴华静 林艳 高鹏 王雯涛 阿晓荟 晏玲 LI Jiatian;WU Huajing;LIN Yan;GAO Peng;WANG Wentao;A Xiaohui;YAN Ling(Faculty of Land Resources Engineering,Kunming University of Science and Technology,Kunming 650093,China;Institute of Information Technology and Cyber Security,Peoples’Public Security University of China,Beijing 100038,China)

机构地区：[1]昆明理工大学国土资源工程学院,云南昆明650093 [2]中国人民公安大学信息技术与网络安全学院,北京100038

出　　处：《武汉大学学报（信息科学版）》2021年第6期816-824,共9页Geomatics and Information Science of Wuhan University

基　　金：国家自然科学基金(41561082)。

摘　　要：模糊航空图像复原不仅能够改善图像的细节与特征,而且可以提高目标的识别能力以及定位精度。在无参数条件下利用已有航空图像建立模糊核估计模型,提出了顾及模糊核连通性的无人机图像半盲复原方法。首先建立梯度筛选,筛选出模糊图像与已有清晰图像梯度域的公共地物,构建保真项;然后利用模糊核梯度八邻域描述模糊核的连通性度量,并将其作为正则项缩小解空间与构建模型;最后重建图像,根据图像金字塔结构对模糊核进行分层估计,并通过分裂Bregman算法解卷积重建图像。对比实验从模糊类型、公共地物、方法对比、方法适用性4个方面进行分析,结果表明,在公共地物达到35%以上时,模糊航片具有良好的复原效果,所提方法具有较强实际应用价值。Objectives:Restoration of blurred aerial image can improve the details and features of images,and enhance the recognition capability and positioning accuracy of targets.Methods:A semi-blind restoration method of unmanned aerial vehicle image considering blurred kernel connectivity is proposed,which uses existing aerial images to establish the model of blurred kernel estimation under the conditions of non-parameters.Firstly,the gradient screening is established to filter out the public features of the blurred image and the existing clear image gradient domain,and construct the fidelity term.Then we use eight neighborhoods of blurred kernel gradient to describe the connectivity measurement of the blurred kernel and use it as a regular term to reduce the solution space and build the model.Finally,we reconstruct the image,estimate the blurred kernel hierarchically according to the image pyramid structure,and reconstruct the image by deconvolution with the split Bregman algorithm.Results:The experiment analyzes and evaluates the proposed method from four aspects:fuzzy type,public ground objects,comparison of methods and applicability of the method.Compared to the existing methods,the experiment results show that,when the public ground objects are over 35%,the blurred aerial images can be restored effectively.Conclusions:Compared with the existing image restoration methods,our proposed method has faster convergence speed,higher accuracy of fuzzy kernel,clearer details and richer information of restored image.

关 键 词：半盲复原 梯度筛选 模糊核连通性 分裂Bregman算法 

分 类 号：P231.1[天文地球—摄影测量与遥感]