# Fit a least squares ellipse

41 views (last 30 days)
B on 21 Apr 2022
Edited: Matt J on 21 Apr 2022
I am trying to fit a least square ellipse to to the 'ellipse' data set that i have attached
I feel like i have everything correct (i might not have) but i am struggling plottting the ellipse function 'f'. i keep getting the following error
Warning: Error updating ImplicitFunctionLine.
Arrays have incompatible sizes for this operation.
Any help would be appreciated, Thanks in advance
%least squares fit to an ellipse
% A x^2 + B xy + C y^2 + Dx + Ey = 1
% Returns coefficients A..F
%A x^2 + B xy + C y^2 + Dx + Ey + F = 0
% where F = -1
M = [x.^2 x.*y y.^2 x y]; % design matrix
b1=ones(size(x));
MTM=M'*M;MTb=M'*b1; %normal equation
R=rref([MTM MTb]); % the coefficent are found in the last collunm
A=R(1:6);
B=R(2,6);
C=R(3,6);
D=R(4,6);
E=R(5,6);
f= @(x,y) A.*x.^2 + B.*x.*y+C.*y.^2+D.*x+E.*y-1; % ellipse function
figure, plot(x, y, '.')
hold on
fimplicit(f)
Alex Sha on 21 Apr 2022
Function: A*x^2 + B*x*y + C*y^2 + D*x + E*y + 1 = 0
Parameter:
A: -0.550741506585047
B: -0.46781995608367
C: -0.804115378029277
D: -0.163329836528614
E: 2.72555198327659

KSSV on 21 Apr 2022
Edited: KSSV on 21 Apr 2022
Replace this line:
A=R(1:6);
with
A=R(1,6);
There is a typo error in getting A. A is a vector in your case, you arenot getting the ellipse.

Matt J on 21 Apr 2022
Edited: Matt J on 21 Apr 2022
I feel like i have everything correct (i might not have)
Your ordinary least squares method is not ideal because your design matrix has stochastic errors in it (iterative total least squares is the gold standard). The method used by this toolbox is a bit better,
and also has utility functions to let you plot directly,
fitobj=ellipticalFit([x,y]')
fitobj =
ellipticalFit with properties:
center: [-0.9975 1.9968]
a: 3.0246
b: 1.9954
angle: -30.0868
plot(fitobj)

R2022a

### Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!