fitting an ellipse to my binary iris image ?

I want to fit an ellipse to the boundary of the iris in my edge map eye image in order to segment the iris from the background.i really appretiate your help. the image is :
thanks inadvance.
na

 Accepted Answer

function report=ellipsefit(XY)
%ELLIPSEFIT - form 2D ellipse fit to given x,y data
%
% report=ellipsefit(XY)
%
%in:
%
% XY: Input matrix of 2D coordinates to be fit. Each column XY(:,i) is [xi;yi]
%
%out: Finds the ellipse fitting the input data parametrized both as
% A*x^2+B*x*y C*y^2+D*x+E*y=1 and [x-x0,y-y0]*Q*[x-x0;y-y0]=1
%
% report: a structure output with the following fields
%
% report.Q: the matrix Q
% report.d: the vector [x0,y0]
% report.ABCDE: the vector [A,B,C,D,E]
% report.AxesDiams: The minor and major ellipse diameters
% report.theta: The counter-clockwise rotation of the ellipse.
%
%NOTE: The code will give errors if the data fit traces out a non-elliptic or
% degenerate conic section.
%
%See also ellipsoidfit
X=XY(1,:).';
Y=XY(2,:).';
M= [X.^2, X.*Y, Y.^2, X, Y, -ones(size(X,1),1)];
[U,S,V]=svd(M,0);
ABCDEF=V(:,end);
if size(ABCDEF,2)>1
error 'Data cannot be fit with unique ellipse'
else
ABCDEF=num2cell(ABCDEF);
end
[A,B,C,D,E,F]=deal(ABCDEF{:});
Q=[A, B/2;B/2 C];
x0=-Q\[D;E]/2;
dd=F+x0'*Q*x0;
Q=Q/dd;
[R,eigen]=eig(Q);
eigen=eigen([1,4]);
if ~all(eigen>=0), error 'Fit produced a non-elliptic conic section'; end
idx=eigen>0;
eigen(idx)=1./eigen(idx);
AxesDiams = 2*sqrt(eigen);
theta=atand(tand(-atan2(R(1),R(2))*180/pi));
report.Q=Q;
report.d=x0(:).';
report.ABCDE=[A, B, C, D, E]/F;
report.AxesDiams=sort(AxesDiams(:)).';
report.theta=theta;

13 Comments

thank you for your answer, but please what is the input parameters (XY)??
The sample x,y coordinates to which you are trying to fit the ellipse.
Sure. Once you obtain the boundary xy coordinates.
Matt J,
Can you use the mango picture, do a example to me to see because i am begginer, i no understand how to obtain the boundary xy coordinates.
Matt J
Matt J on 28 Feb 2013
Edited: Matt J on 28 Feb 2013
That appears to be very challenging! In any case, it's not relevant to this thread or my code, which is about ellipse fitting. It's more of a segmentation problem.
First get rid of the clutter by extracting just the largest object. Then call bwboundaries() to get the perimeter coordinates. Then use the ellipsefit() code.
Image Analyst ,
call bwboundaries like this http://snag.gy/gxOak.jpg ?
Well I can't tell from that image if you called bwboundaries() or not, but what I can tell is that you didn't extract the largest blob into it's own image. That is probably a separate topic, more general purpose than this, and probably should be a separate, new question because lots of people might benefit from it. So, ask a new question like "How can I extract the largest blob in a binary image?"
Matt J,
how to read the image i want using your code or other code because i try to run but cannot read the image. Like below
imread("abc.jpg") a= imread("abc.jpg") imshow(a)
already run like the top code but cannot read the image. i want ask got other code to read the image?
Matt J, how to read the image i want using your code
You were probably really talking to ImageAnalyst.
Except that I didn't post any code.

Sign in to comment.

More Answers (0)

Categories

Asked:

on 12 Nov 2012

Commented:

on 31 Oct 2016

Community Treasure Hunt

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

Start Hunting!