Fit Ellipse to structure in image

I have an image and a reference with many structures similar to below and I want to fit an ellipse on each strcuture such as the one below (in a single image). Once I have a ellipese on all of them , I want to use the ellipse on the difference image (image - reference) and do some computations. How can this segmentation be done in matlab.

Below is a sample image taken from

https://blogs.mathworks.com/steve/2006/06/02/cell-segmentation/

Answer 1

[Matt J]

I recommend downloading this,

https://www.mathworks.com/matlabcentral/fileexchange/87584-object-oriented-tools-to-fit-plot-conics-and-quadrics

IM=im2gray(imread('https://www.mathworks.com/matlabcentral/answers/uploaded_files/1830257/image.png'));
BW=imbinarize(IM);
B=bwboundaries(BW,CoordinateOrder="xy");

fitobj=ellipticalFit(B{1}')

fitobj = 
  ellipticalFit with properties:

    center: [143.6913 104.6159]
         a: 58.7019
         b: 49.6830
     angle: 65.6680

imshow(IM);hold on
plot(fitobj); hold off

●

[Matt J]

If you want to exclude the small protrusions from the main lobe from the fit, you can also download,

https://www.mathworks.com/matlabcentral/fileexchange/90762-further-tools-for-analyzing-objects-in-n-dimensional-images

IM=im2gray(imread('https://www.mathworks.com/matlabcentral/answers/uploaded_files/1830257/image.png'));
BW=~bwlalphaclose(~imbinarize(IM),15);
B=bwboundaries(BW,CoordinateOrder="xy");

fitobj=ellipticalFit(B{1}')

fitobj = 
  ellipticalFit with properties:

    center: [146.5065 106.9114]
         a: 55.9204
         b: 46.0753
     angle: -85.1879

imshow(IM);hold on
plot(fitobj); hold off

●

Answer 2

[Image Analyst]

See the FAQ:

https://matlab.fandom.com/wiki/FAQ#How_can_I_fit_an_ellipse_or_other_shape_to_a_set_of_XY_data?

or see the attached function. You can get the (x,y) coordinates, once you have segmented the gray scale image into a binary image like this:

boundaries = bwboundaries(mask, CoordinateOrder="xy");
%visboundaries(boundaries); % Optional, if you'd like to see them overlaid on the image.
xy = boundaries{1};