vector median filter

Question

Yogesh Joshi 2011-12-28

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/24827-vector-median-filter

采纳的回答： Chandra Kurniawan

is there any function in matlab for vector median filter or vector directional filter? plz help.

2 个评论
显示无隐藏无

Walter Roberson 2011-12-28

Is there a difference between "vector median filter" and "median filter applied to a vector" ?

Yogesh Joshi 2011-12-28

yes, vector median filter is used for color image enhancement.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Chandra Kurniawan 2011-12-28

2
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/24827-vector-median-filter#answer_32632

在 MATLAB Online 中打开

for 1D median filtering, Matlab has command medfilt1

Eq:

medfilt1([4 3 5 2 8 9 1],3)
ans =
     3     4     3     5     8     8     1

4 个评论
显示 2更早的评论隐藏 2更早的评论

Walter Roberson 2011-12-28

color images are 3D matrices.

Chandra Kurniawan 2011-12-29

Yes, I meant 3D matrix

That right function medfilt2 works on 2D image array.

Then, you can use the same function to filter rgb <color> image plane by plane.

I mean you need to apply this to red plane, green plane, blue plane one-by-one.

Here I give you sample code :

I = imread('peppers.png');

J = imnoise(I,'salt & pepper');

imshow(J);

K(:,:,1) = medfilt2(J(:,:,1));

K(:,:,2) = medfilt2(J(:,:,2));

K(:,:,3) = medfilt2(J(:,:,3));

figure, imshow(K);

请先登录，再进行评论。

Answer 2

Image Analyst 2011-12-28

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/24827-vector-median-filter#answer_32671

Whether applying medfilt2() to each color plane independently (such as my code here: http://www.mathworks.com/matlabcentral/answers/16350-how-to-remove-pixels-in-an-image) is an improvement is something you'll have to decide as this can give color artifacts that can make it worse. Maybe you want to convert to HSV and just use medfilt2 on the V channel. Or you can do more sophisticated, smarter things like median filtering the image (somehow) and then replace only SOME of the pixels - the "bad" ones, which you've identified somehow - rather than all of them.