set elliptical ROI to zero in an image

I'm using an algorithm to detect ellipses in an image, it returns the center major minor axis and the orientation, but I would like to delete the detected ones to avoid double detection. I believe the way to do it is to create a mask with an elliptical ROI set it to zero and then subtract it from the original image... anyone knows what function to use? tired to use imellipse but I couldn't figure out a way to get it to work with the detection algorithm.

Answer 1

[Image Analyst]

To erase your image where you have an elliptical mask, do

grayImage(mask) = 0;

See attached demo.

If you need more help, you'll have to post your existing code.

[Mohamed Amine Henchir]

so I believe the code you attached is for the user to define the ellipse. I'm using a function that looks for ellipses in the picture, so the ellipse is not user-defined bestFits = ellipseDetection(E, params); returns an array with 6 values, x and y position of the center of the ellipse, major and minor axis, ellipse orientation and confidence level. so my question is: is there a way to mask the detected ellipses once I have the information given by the EllipseDetection function. thank you!

[Image Analyst]

You can use the FAQ to create an ellipse mask that way: https://matlab.wikia.com/wiki/FAQ#How_do_I_create_an_ellipse.3F

[Mohamed Amine Henchir]

this doesn't allow you to change the orientation of the ellipse...

[Image Analyst]

You can use the rotation matrix (see Wikipedia) like in my attached demo.

[Mohamed Amine Henchir]

thank you very much!

[Mohamed Amine Henchir]

One last question, not sure if I should post it in a different inquiry, anyway so I figured out how to get the ellipses including the rotated one, now I'm trying to use createMask function to create the actual mask, but the method you shared with me, creates the ellipse from x and y points

x = xAmplitude * sind(t + phaseShift);
y = yAmplitude * cosd(t);

Is there a way to get those arrays and create the mask? because createMask only gets for example,

i
thisMask = createMask(gca,[55 10 120 120]);

[Image Analyst]

You can use poly2mask():

mask = poly2mask(x, y, rows, columns);

[Mohamed Amine Henchir]

clc;    % Clear the command window.
workspace;  % Make sure the workspace panel is showing.
clearvars;
format longg;
format compact;
fontSize = 20;
darkGreen = [0, 0.6, 0];
A = imread('one.jpg');
figure;
imshow(A);
hold on;
% Parameterize the equation.
t = linspace(0, 360,1000);
phaseShift = 0;
xAmplitude = 2;
yAmplitude = 1;
x = xAmplitude * sind(t + phaseShift);
y = yAmplitude * cosd(t);
mask = poly2mask(x, y, 256, 256);
imshow(mask);

I attached the image of the mask I get when I show the mask on top of the image, the mask is rectangular shaped which doesn't make any sense, I checked the x,y coordinates they're indeed elliptical...

[Image Analyst]

The mask is a binary image that is true where you have coordinates, and false where there are no coordinates.

Sounds like what you want is to simply plot the ellipse in the overlay with plot, like

hold on;
plot(x, y, 'r-', 'LineWidth', 2);

[Mohamed Amine Henchir]

I would like to delete the detected ellipses from the original image to avoid multiple detection of the same ellipse

[Image Analyst]

How did you detect the ellipses? Give code for that.

[Mohamed Amine Henchir]

I used this function, https://www.mathworks.com/matlabcentral/fileexchange/33970-ellipse-detection-using-1d-hough-transform

if true
  % for i=minA:10:maxA
  params.minMajorAxis = i;
  params.maxMajorAxis = i+5;
  params.numBest = 1;
  params.minAspectRatio = minR;
  params.uniformweights = true;
  params.randomize = 1;
  %params.smoothStddev = 10;
  % note that the edge (or gradient) image is used
  bestFits = ellipseDetection(E, params);
  total = [total; bestFits];
  %fprintf('Output %d best fits.\n', size(bestFits,1));
end
end

[Image Analyst]

What do you mean by "delete the detected ellipses from the original image"? Do you want to blacken the image wherever there is an ellipse? If so, you can use poly2mask() if you have the ellipse outline coordinates. Then do

grayImage(mask) = 0;

[Mohamed Amine Henchir]

here is what I have, for now, I'm using the coordinates of a random ellipse and using poly2mask to create the mask and then subtracting it from the original image

if true
  A = imread('one.jpg');
figure;
imshow(A);
hold on;
% Parameterize the equation.
t = linspace(0, 360,1000);
phaseShift = 0;
xAmplitude = 2;
yAmplitude = 1;
x = xAmplitude * sind(t + phaseShift);
y = yAmplitude * cosd(t);
mask = poly2mask(x, y, 256, 256);
A(mask)= 0;
imshow(A);
end
but nothing is happening!

[Image Analyst]

Your x and y only went from 0-1 - less than a pixel! And your ellipse didn't have a center specified. Try this:

grayImage = imread('cameraman.tif');
subplot(2, 2, 1);
imshow(grayImage);
[rows, columns, numberOfColorChannels] = size(grayImage);
hold on;
% Parameterize the equation.
t = linspace(0, 360, 1000);
phaseShift = 0;
xAmplitude = columns/3;
yAmplitude = rows/5;
xCenter = columns/2;
yCenter = rows/2;
x = xAmplitude * sind(t + phaseShift) + xCenter;
y = yAmplitude * cosd(t) + yCenter;
mask = poly2mask(x, y, rows, columns);
subplot(2, 2, 2);
imshow(mask);
grayImage(mask)= 0;
subplot(2, 2, 3);
imshow(grayImage);