基于Canny边缘检测的半盲图像复原算法  

Semi-blind image restoration based on Canny edge detection

作　　者：孙明竹[1] 赵新[1] 卢桂章[1] 

机构地区：[1]南开大学机器人与信息自动化研究所,天津300071

出　　处：《高技术通讯》2008年第6期602-608,共7页Chinese High Technology Letters

基　　金：国家自然科学基金(60475038);教育部'新世纪优秀人才支持计划'(NCET-07-0464);高等学校博士学科点专项科研基金(20060055037)

摘　　要：将成像系统的点扩散函数限定为二维高斯函数,在此基础上提出了一种新的半盲图像复原算法。该方法将 Canny 边缘检测引入图像复原领域,在选代过程中,首先对当前图像进行边缘检测,并根据检测结果更新模型参数和复原结果。由于在复原过程中将图像边缘与其他区域进行了不同处理,该方法可以在保持图像细节的同时抑制噪声。从仿真实验和实际显微图像复原两方面对此方法进行了验证,实验结果表明,该方法即使在高噪声级别下,仍能得到较好的复原结果。The paper restricts the point spread function of an imaging system to 2D Gaussian function, and on the basis of this, puts forward a new type of semi-blind image restoration algorithm. This approach, which introduces the Canny edge detection into the image restoration, detects edge of the image firstly, and then updates the model parameters and the restoration results in iteration. This approach can preserve image details, and simultaneously, suppress noise during restoration, since edge and region of the image are treated differently. Its. good performance in image restoration was proved with the simulation experiment on digital blurred images and real image restoration, even under the presence of high noise level.

关 键 词：图像复原 边缘检测 各向异性扩散 高斯函数 交替最小化算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TN919.81[自动化与计算机技术—计算机科学与技术]