pdf

一种改进的自适应中值滤波算法

  • 1星
  • 日期: 2018-05-18
  • 大小: 327.07KB
  • 所需积分:1分
  • 下载次数:0
  • favicon收藏
  • rep举报
  • free评论
标签: RAMF

RAMF

滤波算法

滤波算法

在图像的平滑处理过程中  ,如何在噪声滤除的同时保护图像的细节一直是人们研究的热点问题。针对这  一问题  ,在  Hwang和  Haddad提出的一种自适应中值滤波算法  (  ranked2order  based  adap  tive  median  filter,  RAMF)的  基础上  ,提出了一种改进的自适应中值滤波算法。

文档内容节选

2 2 2 2 2 Vol 12 No 7 July 2007 102600 Journal of Image and Graphics 1 2 150080 2 Hwang Haddad ranked order based adap tive median filter RAMF TP391 41 A 3 RAMF 1006 8961 2007 07 1185 04 12 2007 7 7 1 An Im proved M ethod of Adaptive M ed ian F ilter 1 D epa rtm en t of College of M a ths and Com puter Harbin N orm al U niversity Ha rbin 150080 2 College of Inform ation M echanica l Engineering B eijing Institute of Graph ic Comm unication B eijing 102600 GUO Hai xia1 X IE Kai2......

2 2 2 2 , 2 Vol. 12, No. 7 July, 2007 102600) Journal of Image and Graphics 1) 2) , 150080) 2) ( , Hwang Haddad ( ranked order based adap tive median filter, RAMF) , : TP391. 41 : A 3 ; RAMF : 1006 : 8961 (2007) 07 1185 04 12 2007 7 7 1) ( , , ; An Im proved M ethod of Adaptive M ed ian F ilter 1) (D epa rtm en t of College of M a ths and Com puter, Harbin N orm al U niversity, Ha rbin 150080) 2) ( College of Inform ation & M echanica l Engineering, B eijing Institute of Graph ic Comm unication, B eijing 102600) GUO Hai xia1) , X IE Kai2) Abstract To reserve fringe detail of an image while filtering noise in the smoothing p rocess of image still remains a order challenge . In this paper an imp roved adap tive median filter algorithm based on the standard median filter, based adap tive median filter ( RAMF) exam ine for dubious noisy p ixels; noise p ixels within the filter window with high density are filtered according to signal p ixels; within the filter window for noise p ixels with low density are filtered according to fringe detail of the image. W e concluded based on the results of the computer simulation experiments on adap tive median filter that the method p roposed in this paper is better than the standard median filter and RAM F. Keywords is p resented. This method introduces three imp roved aspects: double adap tive, median filter, the Ranked image p rocessing, double exam ine for dubious noisy p ixels, RAMF 1 , : , , , , , Haddad order Hwang ( Ranked RAMF) [ 1 ] , , , based adap tive median filter, , , , 2 , RAMF [ 2 ] , (10551115) : 2006 05 18; : 2006 07 06 : ( 1976 ) , , 163. com E mail: ghx3831350 @ 1186 2 , ( x, y) p N (N p W x, y W x, y W x, y 1 p W x, y Nm ax Ix, y IW m in IW m ax IW m ed RAMF ; : : N =Nm ax , N N ; 3) ; ; ; ; ; , 3 , , , , IW med 2 ; , , : p ; p ; 1 2 med - IW T2 = IW m ed - IW max : T1 = IW T1 > 0, m in T2 < 0, , N 0, IW , m ed , IW m in IW m ax Ix, y = IW m in Ix, y = IW m ax (1) ( (2) : (1) Nmax IW max > IW m ed > IW m in , , , , IW m in IW max , , , , p ) , , , , , Ix, y RAMF , RAMF , ; ( 2 ) 8 1 M SE M AE Tab. 1 Com par ison for M SE, M AE and run tim e to d ifferen t m ed ian f ilters 7 3 3 3 7 3 3 3 7 7 7 7 7 7 11 11 9 9 MSE 0. 007 8 0. 004 5 0. 003 7 0. 003 8 MAE 0. 092 4 0. 069 8 0. 025 6 0. 027 4 12 , , ; ( 3) , , , , , (m s) 547 651 754 679 : (1) 2 (M x, y , Ix, y - IM m in > 0, M x, y p , , RAMF , Ix, y - IW m in > 0 Ix, y - IW max < 0 , M x, y M M p p M >Nm ax ) , Ix, y - IM m ax < 0 ( IM Ix, y M x, y IM m ax m in ) , : , M x, y , , M , M , , M Nmax 4 , Nmax M , , 1 (2) 2 1 [ 3 ] 7 , : : , 5, 0 , 8 ; , 1 : ; , p 8 , , , 1 , N , ( 3) , ? p Ix, y - IW m ax < 0 p p , , Ix, y - IM m ax < 0 , Ix, y; , M x, y j = IM m ax Ii, : j = IM Ii, M x, y ) , m in , n p , : 0 5, ; , Ix, y - IW m in > 0, Ix, y; M x, y m in > 0, Ix, y - IM p , , p : ( p 1187 2 , ; , p , , , M x, y , M x, y 2 med - IW : T1 = IW T1 > 0, T2 = IW med - IW max m in T2 < 0, , N 0, , M x, y Ix, y; , 8 p , k p k ( 1 m k , 8) k p , G k ; [ 4 ] Ix, y : , p1 , p2 , , pn , D k = ( Iik, jk - m k ) 2 ( pk G k ) ( 3) 8 [ 5 ] 1 p dm = dm ( 1 m n) , ( x - xm ) 2 + ( y - ym ) 2 Ix, y = M x, y , RAMF Ixm , ym / dm 1 / dm M x, y dm , , (1) (2) 4 , 1 p 8 Fig. 1 Eight gray line windows for p 2 2 2 , RAMF , 2 Lena 2 ( a) ; 2 ( c) ; 2 ( d) 2 ( e) , , 3; 3, 3 2 ( f) , 3, 3 3 2 ; 2 , 256 2 256 2 12 1 8bits 2 2 ( b) 7 7 7 7, 7 7, 11 11; 7 7, 9 9 2 2 ; : M S E = M A E = 1188 MSE) 2 (mean square error, (mean absolute error, MAE) [ 5 ] ( Ix, y Ix, y - Fx, y ) 2 I2 x, y - Fx, y (4) (5) , Fx, y ( x, y) h 1 ( Ix, y , 0 < x
更多简介内容

推荐帖子

评论

登录/注册

意见反馈

求资源

回顶部

About Us 关于我们 客户服务 联系方式 器件索引 网站地图 最新更新 手机版 版权声明

北京市海淀区知春路23号集成电路设计园量子银座1305 电话:(010)82350740 邮编:100191

电子工程世界版权所有 京ICP证060456号 京ICP备10001474号-1 电信业务审批[2006]字第258号函 京公网安备 11010802033920号 Copyright © 2005-2021 EEWORLD.com.cn, Inc. All rights reserved
$(function(){ var appid = $(".select li a").data("channel"); $(".select li a").click(function(){ var appid = $(this).data("channel"); $('.select dt').html($(this).html()); $('#channel').val(appid); }) })
×