基于相似像素选择的双边滤波图像去噪  

Similar Pixel Selection Based on Bilateral Filter for Image Denoising

作　　者：肖秀春[1,2] 谢仕义[1] 张雨浓[2] 姜孝华[2] 卢晓敏[2] 

机构地区：[1]广东海洋大学信息学院,广东湛江524025 [2]中山大学信息科学与技术学院,广州510275

出　　处：《广东海洋大学学报》2010年第4期54-56,共3页Journal of Guangdong Ocean University

基　　金：国家自然科学基金(60775050);浙江大学CAD/CG国家重点实验室开放课题(A0908)

摘　　要：针对双边滤波在一定程度上模糊了图像特征的缺陷,提出一种基于相似像素选择的图像双边滤波算法。该算法首先在图像中搜索所处理像素的邻域内所有相似像素(与其像素值相近的像素),然后依据相似像素占全部邻域像素比例,确定该像素为特征像素或非特征像素。对于特征像素,采用其相似像素进行双边滤波;对于非特征像素,采用全部邻域像素进行双边滤波。仿真实验表明,该算法相对一般双边滤波,能够在去除噪声的同时,更好地保持图像特征。To overcome drawback of bilateral filter that blurs the image feature to some extent,a bilateral image denoising method based on similar pixel selecting is proposed.For the pixel being processed,the pixels having pixel value are close to this pixel（called similar pixel） within its neighborhood in this image firstly.Then,according to the ratio of the selected similar pixels to all the neighborhood pixels,determine whether conduct bilateral filtering to the similar pixels or to all the pixels within the neighborhood.Simulation results prove that,compared with traditional bilateral filtering,the presented algorithm can remove noise and maintain image feature more effectively at the same time.

关 键 词：图像去噪 双边滤波 相似像素 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]