基于梯度倒频和全变分的运动模糊图像复原  

Motion blurred image restoration based on gradient cepstrum and total variation

作　　者：李俭兵[1,2] 马忍 王志锋[1] 

机构地区：[1]重庆邮电大学通信新技术应用研究所,重庆400065 [2]重庆信科设计有限公司,重庆400065

出　　处：《电视技术》2016年第2期54-58,共5页Video Engineering

摘　　要：通过分析运动模糊图像梯度和Canny算子特性来估算点扩算函数(PSF),提出一种基于结合Canny算子和梯度倒频的运动模糊参数先验估计方法。首先,分析运动模糊图像的梯度信息和Canny算子特性,通过适当的阈值处理后,二者信息结合,利用倒频谱的特性计算出运动模糊尺度和运动模糊方向。最后,用改进的基于半二次型交替数值算法的全变分正则化复原方法对运动模糊图像进行复原。从实验结果可以看出,相对于传统算法,该方法能够较准确地估算出PSF参数,实现比较好的复原效果。By analyzing the motion blurred image gradient and Canny operator characteristics to estimate the point spread func- tion （PSF） , a improved priori estimation method is proposed based on combining Canny operator and gradient cepstrum to es- timate the motion blur parameters. Firstly, motion blurred image gradient information and Canny operator characteristics, through a appropriate threshold, combining both information to calculate motion blur scale and motion blur direction. Finally, using a improved total variation regularization restoration method based on alternating half quadratic algorithms to achieve im- age restoration. From the experimental results, compared with the traditional method, this method can more accurately esti- mate the motion blur parameters and achieve better restoration effect.

关 键 词：运动模糊 退化模型 倒频 梯度 CANNY算子 半二次型交替算法 全变分正则化 

分 类 号：TN911.73[电子电信—通信与信息系统]