基于低分辨率图像约束的BTV图像去模糊算法  

An Improved BTV-Based Image Deblurring Algorithm Based on a Low-Resolution Image Constraint

作　　者：路陆[1] 金伟其[1] 王霞[1] 顿雄 田莉[1] LU Lu JIN Wei-qi WANG Xia DUN Xiong TIAN Li(Key Laboratory of Photoelectronic Imaging Technology and System, Ministry of Education of China, Beijing Institute of Technology, Beijing 100081, China)

机构地区：[1]北京理工大学光电成像技术与系统教育部重点实验室,北京100081

出　　处：《北京理工大学学报》2017年第6期644-649,655,共7页Transactions of Beijing Institute of Technology

基　　金：国家自然科学基金资助项目(61231014;61575023);国家部委预研基金资助项目(40405030302)

摘　　要：多帧图像超分辨率重建算法为了简化算法复杂度,将整个图像重建过程分为数据融合与图像去模糊两步.然而,数据融合过程会丢失一些弱小细节信号,直接使用常规图像去模糊方法是无法将其复原的.为此,提出将低分辨率图像约束(LRIC)引入基于双边全变分(BTV)的图像去模糊优化,并利用梯度下降法求解,获得了BTVLRIC算法.实验表明,对于不同图像内容或数据融合算法生成的数据融合图像,BTV-LRIC法获得的复原图像在视觉效果和客观评价上均优于TV法和BTV法.Multi-frame image super-resolution reconstruction （SRR） algorithms are typically divided into two steps, data fusion and image deblurring, in order to reduce computational complexity. However, some small or weak detail signals lost in the data fusion step cannot be recovered by the conventional image deblurring method. Therefore, a low resolution image constraint （LRIC） was introduced into the traditional deblurring optimization based on the bilateral total variation （BTV） regularization, and then a new deblurring method named BTV- LRIC was obtained in the LRIC based deblurring optimization using gradient descent method. The experiments show that, for data fusion images with different image contents or obtained using different data fusion methods, BTV-LRIC is superior to the TV and BTV method in terms of both visual perception and objective scores.

关 键 词：图像去模糊 图像超分辨 双边全变分 

分 类 号：TN911[电子电信—通信与信息系统]