M = gaussmat(SIGMA) returns a 2D normalized Gaussian of width SIGMA.
The size N of the matrix M is chosen by default as N=CEIL(SIGMA*2)*2+1.
The matrix M is normalized, such that SUM(SUM(M))=1.
M = gaussmat(SIGMA, N) does the same, but constrains the size of the
matrix M to be N*N. Typical values for N are around 3 to 10 times
SIGMA. It is better to choose an odd value for N, so that the central
matrix element is the maximum of the Gaussian.
gaussmat(0) returns 1.
Examples: imagesc(gaussmat(3));
s=vec2scal(loadvec('B00001.vec'),'norm');
s.w=conv2(s.w,gaussmat(1.3),'same');
showscal(s);
F. Moisy
Revision: 1.05, Date: 2005/09/03
See Also
filterf, addnoisef, CONV2.