基于噪声检测的彩色图象脉冲噪声滤波  

Noise Detection-Based Color Filters for Impulse Noise Removal

作　　者：郁伯康[1] 郁梅[1] 蒋刚毅[2] 郑义[2] 

机构地区：[1]杭州电子工业学院电子工程系,杭州310037 [2]汕头大学人工智能与模式识别实验室

出　　处：《计算机工程与应用》2000年第3期55-57,74,共4页Computer Engineering and Applications

基　　金：浙江省自然科学基金!No.698064;广东省自然科学基金!No.980445

摘　　要：文章提出了具有细节保持能力的自适应彩色图像脉冲噪声滤波器，称为细节保持滤波器。新方法对图像中噪声像素进行检测，仅对噪声像素进行有序滤波而对非噪声像素则保持其原值不变，并根据图像噪声情况自适应地选择滤波窗口。从而，有效地滤除随机彩色脉冲噪声、保持图像边缘与细节，其性能优于经典的矢量中值滤波器（ＶＭＦ）、方向一距离滤波器（ＤＤＦ）、距离一幅度矢量滤波器（ＤＭＶＦ）等非线性滤波器。In this paper,new filters for impulse noise removal in color image are proposed based on noise detection, called as detail-preserving filters(DPF).In the new method, impulse noise in color image is first preperly detected,and the detected noisy pixels are filtered out with adaptive selection of filter window,while the noise-free pixels are kept without any change.Experimental results show that the proposed detail-preserving filters remove impulse noise and preserve edges and details very well,that perform better than the classical vector median filter (VMF),direction-directance filter(DDF), and distance-magnitude vector filter(DMVF).

关 键 词：彩色图像 脉冲噪声滤波 噪声检测 图像处理 

分 类 号：TB535.2[理学—物理] TP391.41[理学—声学]