一种基于图论的自适应滤波算法  

A self-adaptive filter algorithm based on graph theory

作　　者：汪云飞[1] 毕笃彦[1] 马和平[1] 

机构地区：[1]空军工程大学航空航天工程学院,陕西西安710038

出　　处：《中南大学学报（自然科学版）》2013年第10期4117-4123,共7页Journal of Central South University:Science and Technology

基　　金：国家自然科学基金资助项目(61175029)

摘　　要：针对脉冲噪声特点,提出一种基于图论的自适应滤波(SAFG)算法。该算法首先设置全局阈值标识出所有可能的噪声点;然后,对每一噪声点进行局部操作,通过比较不同滤波窗口的置信度,自适应调整窗口大小,选择合适的置信滤波窗口;最后,采取两步滤波策略,利用窗口中的非噪声点信息对噪声点进行修复。仿真实验结果表明:SAFG算法不仅能有效抑制强脉冲噪声的干扰,而且可较好地保留图像的细节特征,滤波性能均比经典的中值滤波(MF)和维纳滤波算法的强。In light of the characteristics of impulse noise, a novel self-adaptive filter based on graph theory（SAFG） approach was presented. The first step of this algorithm was to identify the impulse noise nodes of image by setting global threshold; Then, each identified node was partially operated. By comparing confidence degrees of different filter windows, the size of window was self-adaptively adjusted, so the filter window for each noise node was chosen. Finally, a two-step filter strategy was adopted, maked use of the un-noise information around noise node in this confidence filter window to restore noise nodes. The results demonstrate that the filter capabilities of the proposed SAFG algorithm can not only effectively suppress the strong impulse noise disturbing, but also preserve the image details well. It is better than traditional median-filter（MF） algorithm and Wiener filter algorithm.

关 键 词：图像滤波 图论 脉冲噪声 置信窗口 置信度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]