how to smooth / refine edges ??

hello all, i have a final image as an output.. as a result of processing image edges got irregular... how can it be possible for me to smooth/refine just the edges not the complete image... actually i was trying to apply gaussian filter on it it smooths the entire image... P.S image type is RGB .. waiting for replies...

Thanks,

[waqas pervaiz]

Answer 1

[Sean de Wolski]

Similar to what Paul is suggesting:

%Standard IPT Image
I = imread('cameraman.tif');
%Its edges
E = edge(I,'canny');
%Dilate the edges
Ed = imdilate(E,strel('disk',2));
%Filtered image
Ifilt = imfilter(I,fspecial('gaussian'));
%Use Ed as logical index into I to and replace with Ifilt
I(Ed) = Ifilt(Ed);

For an RGB image you may first have to repmat() Ed so that it exists in all three dimensions

Ed3 = repmat(Ed,[1 1 3]);

[Image Analyst]

Why use Canny when thresholding and bwperim will work fine and be a lot faster?

[Sean de Wolski]

example...

[waqas pervaiz]

Thanks alot Sean, but i am getting the following errors:

*Error using iptcheckinput Function EDGE expected its first input, I, to be two-dimensional.

Error in edge>parse_inputs (line 547) iptcheckinput(I,{'numeric','logical'},{'nonsparse','2d'},mfilename,'I',1);

Error in edge (line 190) [a,method,thresh,sigma,thinning,H,kx,ky] = parse_inputs(varargin{:});

Error in ed (line 4) E = edge(I,'canny'); *

[Sean de Wolski]

If you change the image to be three dimensional, you'll need to either extract a color plane from it or use rgb2gray

For example

Igray = rgb2gray(I);

or

Ired_channel = I(:,:,1);

Answer 2

[Paul Kelly]

I've never tried this but I would probably approach it in the following way:

convert the image to grayscale RGB = imread(imageFile); I = rgb2gray(RGB);

binarise it (this will be quite tricky) or find another way to isolate the edge you are interested in. For example:

level = graythresh(I);

BW = im2bw(I,level);

find the edges of the binarised image:

edgeImage = edge(BW,'canny');

Then use morphological functions to broaden the edges to a reasonable size

Finally use the edge image as a mask for the smoothing operation

Answer 3

[Pallawi Pallawi]

Image 1 with the gray background was my input image and I removed the gray background with a white one! Now I need to refine the edges ,making it look smoother . I have tried ,gaussian and above suggestions but it does not works.can someone please help me!LOOKING FOR AN URGENT HELP.