I have two CT image . How can I draw multiple ROIs on both image and calculate mean difference between each the corresponding ROIs with matlab ? I've used the 'imrect' or 'imellipse' but this commands creates the Mask which makes the image as binary image then I would have problem with to calculate mean difference .
How to show the images with the ROIs draw on them?
Not very sure about what you want to do with imrect. This is an idea; the way I would do it. You have to get your hands dirty with actual programming instead of GUI, but it's VERY basic stuff, and easy as soon as you understand indexing, which is very nice in MatLab and the thing you should take with you from this answer:
First of you define the size of your ROIs, which can be easily made with a variable
width=20; %or whatever you wish
height=10;
then define the multiple ROIs using their upper left corner for the position
ROI11=Image1(corner1:corner1+width,corner1:corner1+height); %(width and height eventually the other way around, whatever)
ROI12=Image1(corner2:corner2+width,corner2:corner2+height);
%...
ROI21=Image2(corner1:corner1+width,corner1:corner1+height);
ROI22=Image2(corner2:corner2+width,corner2:corner2+height);
%...
and then calculate the mean however you please, like for example:
Mean1=sum(ROI11-ROI21)/length(ROI11(:));
Mean2=sum(ROI11-ROI21)/length(ROI11(:));
%...
or something along those lines.
Give it a try and play a bit with it.
Related
I have an ellipse in the image.After segmentation i got a broken ellipse as shown .which morphological operation is used to get the perfect ellipse
Actual input file is
output obtained is
i tried imopen ,but i will lose lower ellipse like structure .how to close the upper ellipse like structure without losing lower ones
Mask i created is
i want to segment the ellipse like structure.but some of these structures are connected with rectangular like bodies.how to separate it. erode will eliminate small ellipses
If you want to reconnect something with a mathematical morphology operator, do not use an opening (it increases the gap), but a closing (imclose)! The names are explicits.
In you case, you want to reconnect something vertically cut, so use a horizontal structuring element (type segment).
And yes, you have to invert your image, black pixels representing the absence of information.
Usually, for closing gaps, you would need the close operator.
However, since most software assume active pixels are white, you would either need to invert the image, or use the open operator.
On this image, in matlab, the following works well:
imopen(I,ones(32))
This uses a square structuring element. You may want to experiment with other shapes.
Your example also looks like you moved half of the ellipse, as opposed to some process which deleted pixels in the middle. No simple morphological operation can create a perfect ellipse out of the sample image, unless you use the knowledge that multiple components can be moved to re-form the ellipse. If that is the actual case, you can scan connected components and try to match them together.
Is it possible to create labels for conical surfaces in MS-word.
I have a label ready, but it needs to adapt to that the printed label can be pasted on a conical surface (e.g. a coffee cup)
I doubt it. You could try using WordArt: these are predefined shapes, maybe one of those matches what you want to do.
I suspect you'd be better off using a program like Adobe Illustrator, which can convert text to vector images which you can distort any way you like.
The bigger problem is that the label won't fit properly on a shape that is curved in two dimensions: you'll always have folds somewhere.
I've been working on an image segmentation problem and can't seem to get a good idea for my most recent problem.
This is what I have at the moment:
Click here for image. (This is only a generic example.)
Is there a robust algorithm that can automatically discard the right square as not belonging to the group of the other four squares (that I know should always be stacked more or less on top of each other) ?
It can sometimes be the case, that one of the stacked boxes is not found, so there's a gap or that the bogus box is on the left side.
Your input is greatly appreciated.
If you have a way of producing BW images like your example:
s = regionprops(BW, 'centroid');
centroids = cat(1, s.Centroid);
xpos = centroids(:,1); should then be the x-positions of the boxes.
From here you have multiple ways to go, depending on whether you always have just one separated box and one set of grouped boxes or not. For the "one bogus box far away, rest closely grouped" case (away from Matlab, so this is unchecked) you could even do something as simple as:
d = abs(xpos-median(xpos));
bogusbox = centroids(d==max(d),:);
imshow(BW);
hold on;
plot(bogusbox(1),bogusbox(2),'r*');
Making something that's robust for your actual use case which I am assuming doesn't consist of neat boxes is another matter; as suggested in comments, you need some idea of how close together the positioning of your good boxes is, and how separate the bogus box(es) will be.
For example, you could use other regionprops measurements such as 'BoundingBox' or 'Extrema' and define some sort of measurement of how much the boxes overlap in x relative to each other, then group using that (this could be made to work even if you have multiple stacks in an image).
As when you use subplot to display plots side by side in the same window, I would like a similar function that can do the same for multiple figures. This works:
subplot(2,2,1)
and I am looking for something like this:
subfigure(2,2,1)
Does a build-in function with this advantage exist?
You could write one such function yourself. The idea is to get the screen size:
get(0,'ScreenSize')
then divide it up into sub-regions according to the input parameters. You would also need to account for margins in-between.
Another idea:
create an invisible figure (preferably same aspect ratio as the screen)
call subplot inside it
capture the position of the created axis
delete figure
scale the position captured to fit the screen size, and use it to create the actual figures.
I have a question about the values returned by getPosition. Below is my code. It lets the user set 10 points on a given image:
figure ,imshow(im);
colorArray=['y','m','c','r','g','b','w','k','y','m','c'];
pointArray = cell(1,10);
% Construct boundary constraint function
fcn = makeConstrainToRectFcn('impoint',get(gca,'XLim'),get(gca,'YLim'));
for i = 1:10
p = impoint(gca);
% Enforce boundary constraint function using setPositionConstraintFcn
setPositionConstraintFcn(p,fcn);
setColor(p,colorArray(1,i));
pointArray{i}=p;
getPosition(p)
end
When I start to set points on the image I get results like [675.000 538.000], which means that the x part of the coordinate is 675 and the y part is 538, right? This is what the MATLAB documentation says, but since the image is 576*120 (as displayed in the window) this is not logical.
It seemed to me like, somehow, getPosition returns the y coordinate first. I need some clarification on this.
Thanks for help
I just tried running your code in MATLAB 7.8.0 (R2009a) and had no problems with image sizes of either 576-by-120 or 120-by-576 (I was unsure which orientation you were using). If I left click inside the image, it places a new movable point. It did not allow me to place any points outside the image.
One small bug I found was that if you left-click in the image, then drag the mouse pointer outside the image while still holding the left button down, it will place the movable point outside the image and won't display it, displaying a set of coordinates that are not clipped to the axes rectangle.
I'm not sure of what could be your problem. Perhaps it's a bug with whatever MATLAB version you are using. I would suggest either restarting MATLAB, or clearing all variables from the workspace (except for the image data im).
Might be worth checking to see which renderer you are using (Painter or OpenGL), a colleague showed me some wierd behaviour with point picking when using the OpenGL renderer which went away when using the Painter renderer.
Your code uses the Image Processing Toolbox, which I don't have, so this is speculation. The coordinate system is probably set to the figure window (or maybe even the screen), not the image.
To test this, try clicking points outside the image to see if you can find the origin.