模糊图像迭代盲反卷积复原  

Blurred Image Restored by Iterative Blind Deconvolution

作　　者：刘小芳[1] 秦丽娟[1] 李莹[1] 

机构地区：[1]沈阳理工大学信息科学与工程学院,辽宁沈阳110159

出　　处：《沈阳理工大学学报》2015年第5期28-32,39,共6页Journal of Shenyang Ligong University

基　　金：国家自然科学基金资助项目(61203163);国家重点实验室基金资助项目(2013-006);辽宁社科规划基金项目(L13BJY023)

摘　　要：针对超速或换道车辆容易出现的图像模糊问题,提出了基于盲反卷积的去模糊方法。传统的去模糊方法是假定已知模糊参数,而实际的目标图像模糊参数是未知的。所以采用一种盲复原方法,首先估计出模糊点扩散函数PSF,然后进行模糊处理。根据实际采集的图像含有高斯噪声的特点,将常用零均值高斯白噪声作为其噪声模型,根据噪声均值与方差的最小二乘估计改进盲反卷积模型,得出新的恢复模型,进而恢复含噪声的模糊图像。结果表明,该算法较传统算法恢复效果得到明显改善。To deal with the problems caused by the speeding or changing track of the vehi- cle image fuzzy, a fuzzy method based on blind deconvolution is proposed. Traditional fuzzy method is assumed to known fuzzy parameters, and actual target image fuzzy parameter is unknown. So a method of blind restoration is adopted firstly to estimate the point spread function （PSF） fuzzy PSF,then it is blurred. Simulation results show that recovery effect of the algorithm is superior to traditional methods in comparison with different methods by ex- periment. Also, according to the characteristics of actual collection of image with gaussian noise, the common zero mean gaussian white noise is thought as a noise model;according to the noise mean and variance of the least squares estimate improved blind deconvolution model, which concluds that new recovery model is developed and restores the blurred image containin~ noise well.

关 键 词：模糊 盲反卷积 PSF 车辆 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]