Input matrix must be two dimensional

I'm trying to use the code here .

As you can see, the demo sets the points as follows:

   A set of points
  x = [1 2 5 7 9 6 3 8;
       7 6 8 7 5 7 2 4];

Instead of those points, I wanted to use a 2D image. Thus, I have set

x as follows:

x = imread('xyz.png');

When I try to run the demo, I get the following:

Error using fitellipse>parseinputs (line 321)
Input matrix must be two dimensional

Error in fitellipse (line 81)
[x, params] = parseinputs(x, params, varargin{:});

Error in test_main (line 8)
[z, a, b, alpha] = fitellipse(x, 'linear');

Why is that?

Thanks.

[HumanOptics AG]

Did you find the solution ? Because i've the same problem

Answer 1

[Image Analyst]

fitellipse probably expects an N rows by 2 column list of x,y coordinates. The badly-named x is probably a grayscale or color image, and from the sounds of the error message I'd bet it's a 3D array - a color image of dimensions rows-by-columns-by-3 color channels. You need to pass it the x,y coordinates of the points you want to fit an ellipse to, not an entire image.

[med-sweng]

Thanks for your answer. Do you know of any matlab methods through which I can pass an image with an irregular shape, and have as an output the best ellipse fit for that irregular shape?

I came across this post: http://blogs.mathworks.com/steve/2010/07/30/visualizing-regionprops-ellipse-measurements/

But, I need something being used in literature, so I can refer it in my research for instance (i.e; least squares).

I also came across different code in FileExchange, but, yet, didn't find what I'm looking for.

Thanks.

[Image Analyst]

Yes I do. I have the perfect paper for you: http://www.ecse.rpi.edu/homepages/qji/Papers/ellipse_det_icpr02.pdf. But no, I don't have MATLAB code for it, nor have I studied the algorithm so I can't help you with it. You can try contacting the authors.

[med-sweng]

Thanks for your kind replies.