Is any edge detection used in this image, after having applied Gabor filter?

I'm new to matlab.

I'm trying to implement a paper as my project. The images a) and c) are input images, and according to the paper, Gabor filter has been applied, and the images b) and d) are the outputs respectively. I did apply Gabor filter onto the same image, but my ouptput is way different than the one in the paper.

My question is that, the output images seem to have edges already, and it's not a B/W image completely. So, is there any edge detection technique that was used, if at all it is possible to figure that out by looking at the image?

This is the output I got

Since I couldn't get a dataset with similar images, I used a different set of images lateron but I still couldn't get the desired output. If needed, I'll add that as well.

sigma=0.1;
psi=0.2;
gamma=0.1;
n1=1;
lambda=1;
n2=1;
theta=120;

for i=1:n1
    l=lambda(i);
    for j=1:n2
        t=theta(j);
        g1=gabor_fn(sigma,t,l,psi,gamma);
        %subplot_tight(1,n2,j);
        GT=conv2(I,double(g1),'same');
        figure;
        imshow(GT);
        title('Gabor Filter applied');
        axis image;
    end
end

------------------------

function gb=gabor_fn(sigma,theta,lambda,psi,gamma)

sigma_x = sigma;
sigma_y = sigma/gamma;

% Bounding box
nstds = 3;
xmax = max(abs(nstds*sigma_x*cos(theta)),abs(nstds*sigma_y*sin(theta)));
xmax = ceil(max(1,xmax));
ymax = max(abs(nstds*sigma_x*sin(theta)),abs(nstds*sigma_y*cos(theta)));
ymax = ceil(max(1,ymax));
xmin = -xmax; ymin = -ymax;
[x,y] = meshgrid(xmin:xmax,ymin:ymax);

% Rotation 
x_theta=x*cos(theta)+y*sin(theta);
y_theta=-x*sin(theta)+y*cos(theta);

gb= 1/(2*pi*sigma_x *sigma_y) * exp(-.5*(x_theta.^2/sigma_x^2+y_theta.^2/sigma_y^2)).*cos(2*pi/lambda*x_theta+psi);

I removed noise using Wiener filter to the input image.

^This is what I've used.

Also, could you please shed some light on what basis should I use the Gabor filter parameters, to get the desired output? The references on the internet weren't much helpful.

Thanks a lot in advance.

Answer 1

[Image Analyst]

It does appear that in the lung region some sort of edge enhancement method was used. Not sure what though. Looks sort of like a DOG filter (Difference Of Gaussians) but could be almost anything, but is probably a Gabor filter since that's what you said they said they did in their paper. When you ask "is possible to figure that out" I'm not sure that " that" is. What do you want to filter out from the Gabor-filtered images?

[prahlad h]

The first image I posted is the one I'm working on, and it's the same as the thumbnail.

The paper says Gabor filter was applied and b) and d) are the outputs.

The image I attached, is my output. What possibly could I do in order to attain an output image like the one in the thumbnails?

How is it that the output has so many striations and I get only the segmented lungs?

I was asking you if you could figure out if any edge enhancement/detection is used, to which you already replied.

Sorry for a newbie question.

[Image Analyst]

They masked the image to remove the bulk white stuff. Your algorithm did not mask out the surrounding chest area. Mask it if you want it to look like theirs.

[prahlad h]

I'm not sure if I'm able to convey my point to you.

My question is - look at the lungs. The output in the paper, it has a lot of striations in white. The one I get, has only the segmented lungs, without any sort of striations, or any changes from what was there in the original lung image.

My desired output is the segmented lungs, similar to the images b) and d) in the paper. The background can be taken care of.

[Image Analyst]

Well, what can I say? You're obviously not doing the same operations as they did in the paper. I don't know why.

[prahlad h]

Yes, because all the steps are not mentioned in the paper.

From your knowledge, could you possibly think of the techniques that I could use, in order to get the outputs similar to b) and d) ?

Also, could you please shed some light on how, varying the Gabor filter paramters would vary the output?

Thank you so much already for your time.

[Image Analyst]

Sorry, I have not played around with Gabor filters yet.

[prahlad h]

Any possible way I could turn this

to b) or d) ?

[prahlad h]

Image Analyst?!!

[Image Analyst]

Try varying the filter parameters, or else contact the authors for more detailed instructions.