去除图像椒盐噪声的双向预测算法及其改进  

Bi-directional prediction for the removal of salt-and-pepper noise from images and its improvements

作　　者：孟晓锋[1] 刘培[1] 尹忠科[1] 王建英[1] 

机构地区：[1]西南交通大学信息科学与技术学院,四川成都610031

出　　处：《电路与系统学报》2008年第6期47-51,共5页Journal of Circuits and Systems

基　　金：四川省重点科技计划项目(04GG021-020-5;03GG006-005-2);四川省应用基础研究项目(03JY029-048-2;04JY029-059-2;2006J13-114);国家自然科学基金(60772084)

摘　　要：在去除图像噪声的同时,如何避免图像细节信息的损失和边缘的模糊,是图像处理技术中的一个难点。针对灰度图像中存在的椒盐噪声问题,提出了基于双向预测算法的去噪方法。首先根据椒盐噪声的特点,判断图像像素是信号像素还是噪声像素。对于信号像素,保持灰度值不变;对于噪声像素,利用双向预测的方法来确定处理后该像素点的灰度值。针对上述方法中存在的不足之处,又提出了一种改进方案。改进方案在对噪声像素处理时,根据像素之间的相关性和像素本身的性质自适应地确定预测器的预测系数,提高了预测算法的去噪性能。实验结果表明,本文算法具有良好的去噪特性及细节保持特性。One of pivotal issues in image processing is how to eliminate noise as well as to maintain image detail and edge information. To solve the problem of salt-and-Pepper noise in the gray image, a new denoising algorithm is proposed based on bi-directional prediction algorithm. According to characteristics of salt-and-pepper noise, one pixel to be processed is estimated at first to be a signal pixel or a noise pixel. The Signal pixel is kept untouched to preserve the detail information of the image. To the noise pixel, bi-directional Prediction algorithm is employed depending on the surrounding pixels. An improved algorithm is put forward to overcome the disadvantages of the above method.To the noise pixeh according to the correlations of the neighboring pixels and their characteristics, the improved algorithm sets adaptively prediction coefficients, which can improve the above algorithm＇s performance. The experimental results demonstrate that the algorithm presented here is good both at preserving the details of the image and at denoising

关 键 词：图像处理 图像去噪 椒盐噪声 双向预测 中值滤波 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]