roipoly for multiple shapes

I'm wanting to create multiple closed unconnected shapes using roipoly. I can do this fine for 2 shapes but when I add in the code for a 3rd shape it always connects the last coord of the 3rd shape to the first coord of the 1st shape.

if true
xmask_temp = [];
ymask_temp = [];
junk = [];
for i=1
  A=imread(['Image0001.bmp']);
  [junk,xmask_temp,ymask_temp]=roipoly(A);
  a = 1+size(xmask_temp,1);
  xmask(1:size(xmask_temp,1),i)=xmask_temp;
  ymask(1:size(ymask_temp,1),i)=ymask_temp;
  [junk,xmask_temp,ymask_temp]=roipoly(A);
  b = a-1+size(xmask_temp,1);
  xmask(a:b,i)=xmask_temp;
  ymask(a:b,i)=ymask_temp;
  b = b+1;
  [junk,xmask_temp,ymask_temp]=roipoly(A);    % where the problem starts
  c = b-1+size(xmask_temp,1);
  xmask(b:c,i)=xmask_temp;
  ymask(b:c,i)=ymask_temp;
end
B = roipoly(A,xmask,ymask);
imshow(B);

Answer 1

[Image Analyst]

In the loop, call poly2mask() to get a new mask for the new polygon they draw. Then OR it in with the image.

Before the loop

mask = false(rows, columns); % Initialize final mask

Then in the loop

thisMask = poly2mask(xVertices, yVertices, rows, columns);
mask = mask | thisMask; % Combine this mask with the final mask.

That way you build up the final mask by ORing in each individual mask one at a time. Change variable names to whatever you're using, of course.

[Thomas Darwent]

Hi, Thanks for your help. This method does build up an image of unconnected shapes but for the application we are using we need to build this image using roipoly.

[Image Analyst]

Why do you say "but"? You still use roipoly(). You just also use poly2mask to create a mask from those vertices, and then OR in that mask with a cumulative mask you're building.

If you can't figure it out, attach 'Image0001.bmp' and I'll make a full demo.