散焦图像的维纳滤波复原及振铃抑制  

Defocused Image Restoration with Wiener Filter and Ringing Suppression

作　　者：唐满芳 胡宗福[1] 

机构地区：[1]同济大学电子与信息工程学院,上海

出　　处：《图像与信号处理》2015年第4期87-93,共7页Journal of Image and Signal Processing

摘　　要：图像复原是数字图像处理的重要研究内容。基于频域的维纳滤波实现图像复原,并针对复原过程中伴随的振铃效应,采用扩展边缘的方法做了消除振铃预处理。然后进一步分析点扩散函数半径、图像梯度对振铃样式的影响,并由此建立边界振铃样式数据库R。并结合轮廓提取以及边缘膨胀法,得到振铃影响域内某点到轮廓的距离,从而调用对应库里的振铃样式做振铃修正。最后基于Matlab用简单轮廓图形验证该振铃抵消方法的有效性,从而将振铃幅度控制在±2之内,极大程度的恢复了原始高清图像。Image restoration is a key research point of digital image processing. Wiener filter is used for basic image restoration, and boundary extension is a good method of ringing suppression preprocessing. Further analysis is made to find the relation between ringing pattern and precondition, including PSF radius and the gradient of outline. According to the relation, we build a 2D database matrix R to store the outline ringing pattern information. Combined with edge detection and image dilation, we can invoke the pattern database to suppress the ringing in a certain pre-condition and distance. Finally we verify the effectiveness of method with a simple outline example in Matlab, thus controlling the amplitude of ringing within ±2, and then the image information is restored as much as impossible.

关 键 词：图像复原 维纳滤波 点扩散函数 振铃 边缘提取 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]