改进的二值图像复原方法研究  

Study of improved method for bi-level image restoration

作　　者：丁卫[1] 王忠[1] 

机构地区：[1]四川大学电气信息学院,四川成都610064

出　　处：《计算机工程与设计》2011年第7期2405-2408,共4页Computer Engineering and Design

基　　金：国家自然科学基金项目(49901013);中国博士后基金项目(2005038615);四川科技厅应用基础基金项目(2008JY0043-1)

摘　　要：为了解决传统基于GBB方法的正则化二值图像复原算法放大背景噪声的问题,提出了一种基于GBB方法的改进的正则化二值图像复原算法。根据提出的算法进行实验,定性的对两种算法的图像复原结果进行了比较。同时,利用ISNR选取最佳参数,且在不同程度噪声的影响下定量的评价了两种复原方法的结果。实验结果表明,该方法对于模糊二值图像的复原效果有显著提升,有效降低了噪声对图像复原的影响。To resolve the problem of the amplifying of background noise in the traditional regulafization method for bi-level image restoration based on GBB method, an improved regularization method for bi-level image restoration based on the global Barzilai and Borwein （GBB） method is proposed. According to the improved method, the experiment is implemented to qualitatively analyze the restoration results of the two methods. At the meanwhile, the optimal parameter is selected by ISNR and the results of the two methods in the different degree of noise environments are quantitatively evaluated. Simulation results show that the improved method has a remarkable improvement for the bi-level image restoration and effectively lowers the effects of noises in the bi-level image restoration progress.

关 键 词：二值图像复原 新的正则化方法 GBB方法 阈值化 中值滤波 

分 类 号：TN911.73[电子电信—通信与信息系统]