I need help on gradient mask. Thanks a lot.

%I have following script trying to produce the Gaussian mask Hx and Hy and get the gradient x (Gx) and gradient Y(GY) of image by convolving the mask to the image. Gx = conv2(Image,Hx), Gy = conv2(Image,Hy). but the result looks not correct. Gx looks like gradient y and Gy looks like gradient y. Is anybody familiar with gradient mask and give me any suggestion? thank you very much.

Hx = zeros(13);
Hy = zeros(13);
sigmasqrd = 1/9;
S = (-6:6);
for i = 1:13
    for j = 1:13
        xpo = 2.0*exp(-(S(i).^2+S(j).^2)*sigmasqrd)*sigmasqrd;
        Hx(i,j)= S(i)*xpo;
        Hy(i,j)= S(j)*xpo; 
    end
end

Answer 1

[Image Analyst]

Add this code to the end of your code:

Hx
Hy
subplot(1,2,1);
imshow(Hx, []);
subplot(1,2,2);
imshow(Hy, []);

The code does produce kernels that should provide a gradient-like operation. Does that answer your question? There is also imgradient() in the Image Processing Toolbox if you want to use one of the standard gradient methods.

Answer 2

[Zhongkui]

thank you for your answer. You answered part of my question. my second question is if I reversed the X and Y gradient, i.e. Hx produce the Y gradient while Hy makes X gradient. Thank you again for your answer.

Zhongkui

[Image Analyst]

Yes, it would. What you call the variable does not affect what it does.

[Zhongkui]

thank you.