Gaussian Filter Terminology - sigma, kernal

I am looking at code people have written to make their own filters in Matlab. These calculations below seem to keep popping up. I understand sigma is the total sum of all values in the kernal grid. Correct? But where in this calculation do you specify the grid/kernal size? eg 3x3, 5x5, etc.

I wish to understand this code and can you advise me on how you can change the size of the kernal/grid. Thank you in advance.

    sigma = 1;
    inp = (-1 .*2.5 .*sigma):1:(2.5 .* sigma);  
    gauss1d = (1/(sigma .* sqrt(2*pi))).*exp(-(inp.^2/(2.*sigma.*sigma))); 
    gaussprime = diff(gauss1d); 

    inp =
    -2.5000   -1.5000   -0.5000    0.5000    1.5000    2.5000
    gauss1d =
    0.0175    0.1295    0.3521    0.3521    0.1295    0.0175
    gaussprime =
    0.1120    0.2225         0   -0.2225   -0.1120

Answer 1

[Thorsten]

You got it wrong. Sigma determines the width of the Gaussian. The larger sigma, the more values you need to sample the Gaussian properly.

 function z = gauss(sigma)
 %GAUSS 1-D Gaussian.

   precision = 3; %
   bound_x = ceil(precision * sigma);
   x = (-bound_x : bound_x);

   z = 1/(sqrt(2*pi) * sigma) * exp(-0.5 * (x.^2 / sigma^2));
   z = z/sum(z(:)); % normalize such that the Gaussian integrates to unity

[Sam]

Ok thanks for clarifying that. Using the calculations i found above, how would i change the size of the Gaussain?

[Thorsten]

By changing sigma.

[Sam]

I thought sigma was the value for the standard deviation of the gaussian and the calculation needed another value to determine the actual size?

[Thorsten]

The size is given by inp which depends on sigma.