can any one help me to implement this matlab code

Hi,

I need your help please to implement a matlab code for Edge drawing Edge Drawing works on grayscale images and is comprised of 4 steps:

(1) Suppression of noise by Gaussian filtering,

(2) Computation of the gradient magnitude and edge direction maps,

(3) Extraction of the anchors (peaks of the gradient map),

(4) Linking of the anchors by smart routing to compute the final edge map.

This is a link that can explain the method.

http://ceng.anadolu.edu.tr/CV/EdgeDrawing/

thanks

[Walter Roberson]

show the code you have arrived at so far, and describe the error messages you received when you tried your code.

[Image Analyst]

The link has a download link. Have you tried to download the code, and translate into MATLAB if necessary? Who do you think should do that?

[Walter Roberson]

The download link leads to binaries.

The algorithms in Table 1 and Table 2 would be easily translated to MATLAB.

[Pamela Paolo]

Hi, the code in the download link is programmed in C++. I have no idea about this language. I tried to do it with matlab

I=imread('image.jpg');
figure;imshow(I);  
I2=rgb2gray(I);
figure;imshow(I2);
H = fspecial('gaussian');
I3 = imfilter(I2,H,'replicate');
figure;imshow(I3);
imgrad = gradient(I3)
figure;imshow(imgrad);

But I have this error

??? Error using ==> rdivide Integers can only be combined with integers of the same class, or scalar doubles.

Error in ==> gradient at 75
      g(2:n-1,:) = (f(3:n,:)-f(1:n-2,:))./h(:,ones(p,1));

Error in ==> edgedrawing at 8
imgrad = gradient(I3)

Also i couldn't implement the matlab code of edge direction maps, extraction of the anchors and linking of the anchors by smart routing to compute the final edge map.

[Image Analyst]

Try casting I3 to single before passing it in.

[Pamela Paolo]

thanks for your reply

I=imread('image.png');
figure;imshow(I);
I2=rgb2gray(I);
figure;imshow(I2);
H = fspecial('gaussian');
I3 = imfilter(I2,H,'replicate');
figure;imshow(I3);
I3 = single(I3)
imgrad = gradient(I3);
figure;imshow(imgrad);

But i have this result

with this image

[Thien Thu Ngo]

Hi @Pamela Paolo , could you please share with me the C++ code of this paper? Thanks a lot.

Answer 1

[Pamela Paolo]

What is the purpose of casting I3 to single?? becuse i haven't a good result of gradient

[Image Analyst]

I haven't run your code but when it says " Integers can only be combined with integers of the same class, or scalar doubles." it means that it's trying to do something with your integer array, like combine it somehow with a single or double array, and that is against MATLAB's rules. There is no such rule with singles and doubles combining with each other so that's why I said to cast it to single.

[Pamela Paolo]

Thanks for your reply

Have you an idea about the programming of these steps with matlab:

-edge direction maps

-extraction of the anchors

-linking of the anchors by smart routing to compute the final edge map.

Answer 2

[Pamela Paolo]

Hi

To find the gradient can i use

serl=[1 1 1,1 1 1, 1 1 1 ];
image1=imdilate(I3,serl);
image2=imerode(I3,serl);
grad=image1-image2;

what are the matlab functions used to program these steps:

-edge direction maps

-extraction of the anchors

-linking of the anchors by smart routing to compute the final edge map.

[Image Analyst]

I don't think that would give the gradient as one would normally think about it. There is a gradient() function you know. You could also use conv2(yourImage, [-1 -1 -1; -1 8 -1; -1 -1 -1]); For the rest, it sounds like you have a particular paper in mind, and I imagine that should give the details for each of those steps.

[Pamela Paolo]

Hello

I have only this link http://ceng.anadolu.edu.tr/CV/EdgeDrawing/. I haven't more details. That is why I asked the question. thanks

I=imread('image.png');
figure;imshow(I);
I2=rgb2gray(I);
figure;imshow(I2);
H = fspecial('gaussian');
I3 = imfilter(I2,H,'replicate');
figure;imshow(I3);
I3 = single(I3)
imgrad = gradient(I3);
figure;imshow(imgrad);

[Pamela Paolo]

Hello,

I need your help to understand these steps

-Computation of the gradient magnitude and edge direction maps,

-Extraction of the anchors (peaks of the gradient map),

-Linking of the anchors by smart routing to compute the final edge map.

It's correct to do this to compute the gradient magnitude and edge direction maps?

[Gmag,Gdir] = imgradient(I3)