基于弱表层模型的降晰函数辨识与图像复原  

Blur function identification and image restoration based on a weak membrane model

作　　者：赵剡[1] 张春晓[1] 许东[1] 

机构地区：[1]北京航空航天大学仪器科学与光电工程学院导航系,北京100083

出　　处：《红外与激光工程》2007年第z2期282-285,共4页Infrared and Laser Engineering

摘　　要：针对常规正则化易产生过度平滑解的情况,采用了一种具有保边特性的弱表层模型。该模型因引入反映不连续性边缘的线过程,从而使平滑在边缘处得到有效抑制。对得到的非线性代价函数,采用一种混合退火策略交替迭代估计图像、线场和降晰函数。在确定线元阈值的选取中引入信噪比,使得在每次迭代中阈值根据图像复原情况进行更新,通过仿真合成图像实验,验证了该改进方法的有效性。Blur function and true image are estimated simultaneously based on a special Markov random field model(MRF),termed weak membrane model,combining with Maximum posterior estimation(MAP).Contrast with the standard regularization producing over smoothed solution,thus losing the information of edges,this model inhibits the smoothing implementation at edges due to introducing line process presenting the discontinuities.A mixed-annealing algorithm is used to deal with the non-convex cost function,alternately providing estimations of true image,line process and blur function.The threshold,which determines line elements,is altered depending on the signal to noise ratio of the current restored image in each iteration operation.Experiment results show that the improved selection of the threshold is more efficient than a constant threshold.

关 键 词：图像复原 降晰函数辨识 线过程 弱表层模型 

分 类 号：TN21[电子电信—物理电子学]