The program generates a 2D Gaussian. The program then attempts to fit the data using the MatLab function “lsqcurvefit “ to find the position, orientation and width of the two-dimensional Gaussian.
Execute “mainD2GaussFitRot.m” with not input parameters.
The two-dimensional Gaussian function is defined by the function “D2GaussFunctionRot.m” and “D2GaussFunction.m”
Jan Simon's criticism is misplaced, only minor changes are needed to plot real data. If you have a regular grid of points, all you must do is:
change line 29 to the appropriate initial guesses
change line 30 to xs=x0;
change line 31 to noise=0;
change line 38 to [X,Y] = meshgrid(x,y);
where x and y a vectors with your x an y axis points
change line 47 to Z = zdata;
where zdata is a length(x) by length(y) matrix with the data you wish to fit such that zdata(i,j)=z(x_i,y_j).
Line 48 should be commented out.
If your grid points are not regular, you must set up the X, Y, and Z matrices so that they are all the same size and element i of X, Y and Z all correspond to the point (X(i),Y(i),Z(i)). There might be additional restrictions on the size and shape (I think it must be a matrix and not an array), but I'm not certain.
At any rate, thank you very much Gero. This is very helpful. :)
What a pity that this is a script, which generates the data by a random function. Before it can be used for real problems, substantial changes are required.
It would increase the usability, if the calculations are move to a function and separated from the generation of test data.