Returning cropped image on to original image after doing some analysis on that cropped portion

Returning cropped image on to original image after doing some analysis on that cropped portion. Basically that cropped portion became contour now after doing some analysis, i just wanted to put it back to its original place. I cropped using imcrop, i have cropped image coordinates [Xmin Ymin Width Height]

Can anyone help me out to do this.

Thanks in advance

Answer 1

[Matt J]

i0=floor(Ymin); 
j0=floor(Xmin); 

i1=i0+ceil(Height)-1;
j1=j0+ceil(Width)-1;

Image(i0:i1,j0:j1)=croppedImage  %Put croppedImage back in original location

[Image Analyst]

If you get the cropping box from something like regionprops, it's half a pixel outside the region, so you'd want to use ceil to get i0 and j0, not floor, right?

[Matt J]

I suppose. I have to say I find it bizarre that these functions don't give bounding box specs as i0,j0,i1,j1 directly.

[Image Analyst]

Matt: yes, tell Steve Eddins. I've already put in a suggestion to have an option to return bounding box along actual integer pixel boundaries like most people would expect.

[HARISH KUMAR]

@Matt J Thank you for your respone. I tried your code it is giving some error. Please see attached file.

figure
A = imread(FileNames1{1});
i0=floor(Rect(2)); 
j0=floor(Rect(1)); 
i1=i0+ceil(Rect(4))-1;
j1=j0+ceil(Rect(3))-1;
A(i0:i1, j0:j1) =  'myFigure.tiff' % myFigure.tiff is a contour 

Thanks in adavance

Harish

[Matt J]

@HARISH KUMAR Perhaps you meant,

A(i0:i1, j0:j1) =  imread('myFigure.tiff')

In any case, the right hand side has to be a numeric array the same size as A(i0:i1, j0:j1).

[HARISH KUMAR]

@Matt J Thank you.

As i said we did some analysis on cropped image (for examples 444 -by -288) and it became Contour, now contour dimensions (656-by-875-by-3) has been change (bigger than the cropped image ). Please see attached error file and contour. why it became 3D. Pleae guide me.

A = imread(FileNames1{1});
i0=floor(Rect(2)); 
j0=floor(Rect(1)); 
i1=i0+ceil(Rect(4))-1;
j1=j0+ceil(Rect(3))-1;
A(i0:i1, j0:j1) =  imread('myFigure321.tiff')
figure
imshow(A);

Thanks in advance

[Matt J]

@HARISH KUMAR

Please do not attach screen shots of text. It consumes much more memory. Just copy/paste the text.

Your attachments don't include myFigure321.tiff, so we have no idea what it contains. However, it is likely 3D because it is a color image with RGB channels. You can use im2gray or rgb2gray to convert an RGB image to grayscale. You can also then use imresize to make it the size that you need to reinsert it into the space occupied by A(i0:i1, j0:j1).

[HARISH KUMAR]

@Matt J Okay, i understand.

Thanks for guiding me. myFigure321.tiff is nothing but Contour (which is i have attached in the previour comment ).

I have modified my code according to your directions. But i have to insert color image (Contour) on to original image not grayscale image and that too without white spaces around contour (i can see some white spaces around contour, i dont want those white spaces).

A = imread(FileNames1{1});
i0=floor(Rect(2)); 
j0=floor(Rect(1)); 
i1=i0+ceil(Rect(4))-1;
j1=j0+ceil(Rect(3))-1;
GRAY = im2gray(imread('myFigure321.tiff'));
RESIZE = imresize(GRAY, [size(A(i0:i1, j0:j1))]);
A(i0:i1, j0:j1) = RESIZE;
figure
imshow(A);

I am attaching output of this code (that is on right side of image. i want image (contour on left side to be on the right of the image (not grayscale with white spaces)). Please See attachment you will get an idea.

Thanks for your patience and support

Harish

[Matt J]

To embed as RGB,

A=repmat(A,1,1,3);

RESIZE = imresize(imread('myFigure321.tiff'), [i1,j1]-[i0,j0]+1);

A(i0:i1, j0:j1,:) = RESIZE;

To export the contour plot figure without a white boundary, you need a better export tool, such as export_fig.

[HARISH KUMAR]

@Matt J Thank yo so much, it is working perfectly.

I used tight border and exportgraphics for trimming borders since export_fig is giving error like Unrecognized function or variable 'export_fig'.

Finally i just want to know that how can i put colorbar out side my figure (it should not be on my figure), Please let me know.

Attaching figure for your reference.

Thanks in advance

Harish

[Matt J]

Unrecognized function or variable 'export_fig'.

Probably because you didn't download it?

Finally i just want to know that how can i put colorbar out side my figure (it should not be on my figure), Please let me know.

The colorbar(ax) command should be all that's required, where ax is the axes handle for your overall image.

[HARISH KUMAR]

@Matt J

Hi, Thanks for your response.

i have tried colorbar(ax), problem here is, i am generating contour,saving it and inserting back that contour to its place.

When we are plotting it and saving, it is saving colorbar along with contour by changing size of the my contour to smaller (basically colorbar occupying some space of my contour).

i want that colorbar out of my contour and my main image. Please look at attached image you will get an idea how it is generating.

Thanks in advance

Harish

[Matt J]

When we are plotting it and saving, it is saving colorbar along with contour by changing size of the my contour to smaller

Well, you should be removing the color bar before saving and exporting. That should be easy enough to do with the colorbar('off') command.

But I agree with @DGM that maybe what you are ultimately trying to do can be done entirely by overlaying handle graphics.

[HARISH KUMAR]

@Matt J

Thanks! it is helpful it

then how to turn on colorbar after returning contour to main image.

Thanks in advance

Harish

[Matt J]

The colorbar commandcan be used to both add and remove a color bar from any axes.

[HARISH KUMAR]

@Matt J Thank you

but it is not displaying colorbar even after putting colorbar command after returning saved contour to main image. Please see the code let me know if i place colorbar command in wrong place.

A = imread(FileNames1{1});
i0=floor(Rect(2)); 
j0=floor(Rect(1)); 
i1=i0+ceil(Rect(4))-1;
j1=j0+ceil(Rect(3))-1;
A=repmat(A,1,1,3);
RESIZE = imresize(imread('myFigure314.tif'), [i1,j1]-[i0,j0]+1);
A(i0:i1, j0:j1,:) = RESIZE;




figure
colorbar; % i kept colobar ('off') before saving contour
imshow(A,'border', 'tight');

Thanks in advance

Harish

[Matt J]

figure

imshow(A,'border', 'tight'); colorbar;

[HARISH KUMAR]

@Matt J

Thank you so much. It is working great.

I would appreciate your patience and support.

[HARISH KUMAR]

Hi @Matt J

I just got realised that i can not see my contour background (my original image ) when i reduce the opacity (FaceAlpha value).

I used export_fig. to export my figure.

Can you please help me out to see my original background image when reduce value of FaceAlpha.

Thanks in advance

[Matt J]

There is nothing behing the contour plot, so there's no reason you should see something behind it when you reduce opacity. It sounds more and more like overlaid graphics was what you intended all along.

[HARISH KUMAR]

Thanks for your response,

@Matt J There was a background image on which i did some analysis then it became contour. Now i want to see that background image throgh contour (We do call transparency).

i am attaching image for your refence.

How i wanted to be on left side of image, how i am getting on right side of image.

Thanks in advance

[Matt J]

It looks like you already have the means to make what you want, else where did the left image come from?

[HARISH KUMAR]

Left side image from software application.

[Matt J]

Well, it's the kind of thing you can do as well in Matlab using DGM's answer.

[HARISH KUMAR]

Okay, Thank you.

I will try.

[HARISH KUMAR]

@Matt J Can not possible to get transparency of my contour (Apart from DGM solution) ?

Answer 2

[DGM]

It seems like this is something that would be easier done in-figure, rather than trying to cut up screenshots.

% ROI extents as subscript vectors
roiy = 30:150; 
roix = 50:150;

% get image and sample region
inpict = imread('cameraman.tif'); % 256x256x1

% get the sample region and filter it or something
roisample = inpict(roiy,roix);
roisample = imgaussfilt(roisample,5);

% draw image (must be RGB!)
if size(inpict,3) == 1
    safepict = repmat(inpict,[1 1 3]);
end
imshow(safepict); hold on

% overlay the contourf plot
[~,hc] = contourf(roix,roiy,roisample,10);
hc.LineStyle = 'none';

% if you want the contour plot to appear transparent
% overlay another copy of the image with alpha
%hi = imshow(safepict);
%hi.AlphaData = 0.1;

colormap(parula)
colorbar

[HARISH KUMAR]

@DGM Thanks for response.

your code looks easier i will try to impliment this in my code.

Thank you.

[DGM]

While the prior example works in older versions, if you want to use the recent functionality of contour objects for implementing the transparency:

% ROI extents as subscript vectors
roiy = 30:150; 
roix = 50:150;

% get image and sample region
inpict = imread('cameraman.tif'); % 256x256x1

% get the sample region and filter it or something
roisample = inpict(roiy,roix);
roisample = imgaussfilt(roisample,5);

% draw image (must be RGB!)
if size(inpict,3) == 1
    safepict = repmat(inpict,[1 1 3]);
end
imshow(safepict); hold on

% overlay the contourf plot
[~,hc] = contourf(roix,roiy,roisample,10);
hc.LineStyle = 'none';
hc.FaceAlpha = 0.5; % won't work in anything older than R2022b

colormap(parula)
colorbar

Like I mentioned though, the moment you start mixing it with the background, the colorbar becomes difficult to interpret.

[HARISH KUMAR]

@DGM Thank you so much for update.

I will try to impliment it.

[HARISH KUMAR]

@DGM Your code is working perfectly, thanks for efforts.

Here, you have taken ROI coordinates and plotting contour using ROI coordinates but in my case i will select ROI using imcrop tool.

My Matlab analysis code do analysis (some kind of displacement analysis) on image (cropped ROI), generateS X coordinates, Y coordinates and displacement coordinates on that cropped ROI, using these i will generate contour like contourf(X,Y, displ_cordi). Here how can i give actual ROI Coordinates to overlay that contour on my image.

Thanks in advance!

[DGM]

If you are using imcrop() interactively, then make sure to get the rect output parameter.

% manually crop the image and get the rect parameter
[roisample srect] = imcrop(inpict);

% ...
% do a bunch of other stuff
% ...

% get the integer subscripts that imcrop() uses for selecting the image region
roix = round(srect(1)):round(sum(srect([1 3])));
roiy = round(srect(2)):round(sum(srect([2 4])));

% draw image (must be RGB!)
if size(inpict,3) == 1
    safepict = repmat(inpict,[1 1 3]);
end
imshow(safepict); hold on

% overlay the contourf plot
[~,hc] = contourf(roix,roiy,roisample,10);
hc.LineStyle = 'none';
hc.FaceAlpha = 0.5; % won't work in anything older than R2022b

On the other hand, maybe you're talking about using some coordinate space other than pixels? How are X and Y generated? Can you give a simple example for how you are going from the cropping task to the contour plotting?

[HARISH KUMAR]

Actually, i am modifing existing matlab code to get better output, even though i am new to matlab.

After looking at that code, as per my understanding, he created function, stated some inputs required for that function. In that function, he created mesgrid by using the size of that ROI and some stated inputs. Later some image correlation has happen, from this we got those three components which is used for generating contour.

I have realised that size of 1,2 and 3 components are same like same number of rows and columns.

Using these components i have generated contour ( when i try to overlay this contour on my original image, this contour is going and falling on top left corner of the image (irrespective of ROI))

Where as in your case, size of roix = 1*n, size of roiy = 1*m and size of roisample = n*m

[DGM]

contour() and contourf() both accept X and Y data as either vectors or as a meshgrid (matrix) format.

x = 1:10; % a 1x10 vector
y = 1:20; % a 1x20 vector
[X Y] = meshgrid(x,y); % two 10x20 matrices
Z = x.*y.'; % a 10x20 matrix

subplot(1,2,1)
contourf(x,y,Z) % using vector XData,YData
subplot(1,2,2)
contourf(X,Y,Z) % using matrix XData,YData

What matters is that the image object and the contour object have their XData and YData in comparable units. In order for that to work, either the contourf() object needs to have X and Y in pixels, or the call to imshow() needs to specify XData and YData in whatever coordinates the contourf() plot is using.

In the prior examples, I did everything in image coordinates. Plotting can still be done in image coordinates, even if the Z data isn't calculated in image coordinates. So long as the X and Y vectors/matrices you're feeding to contourf() are compatible in size with the Z data you give to contourf(), the values therein can be in any units.

figure
% ROI extents as subscript vectors (in image coordinates)
roiy = 30:150; 
roix = 50:150;

% get image and sample region
inpict = imread('cameraman.tif'); % 256x256x1

% get the sample region
roisample = inpict(roiy,roix);

% say you're using some arbitrary X,Y to generate the Z data ...
x = linspace(-1.5,1.5,numel(roix));
y = linspace(-0.5,0.5,numel(roiy));
[X Y] = meshgrid(x,y);
roisample = double(roisample).*(X+Y); % some hypothetical operation (X and Y are not in px)

% draw image (must be RGB!)
if size(inpict,3) == 1
    safepict = repmat(inpict,[1 1 3]);
end
imshow(safepict); hold on

% ...the X,Y data used for plotting can still be in px
[~,hc] = contourf(roix,roiy,roisample,10);
hc.LineStyle = 'none';

colormap(parula)
colorbar
axis on % just to show the coordinate system

If instead, you want to plot the image in the same coordinate space that was used to generate the Z data, then that's doable, but I think it's a lot more cumbersome. Note the range of the xticks and yticks.

figure
% ROI extents as subscript vectors (in image coordinates)
roiy = 30:150; 
roix = 50:150;

% get image and sample region
inpict = imread('cameraman.tif'); % 256x256x1

% get the sample region and filter it or something
roisample = inpict(roiy,roix);

% say you're using some arbitrary X,Y to generate the Z data ...
x = linspace(-1.5,1.5,numel(roix));
y = linspace(-0.5,0.5,numel(roiy));
[X Y] = meshgrid(x,y);
roisample = double(roisample).*(X+Y); % some hypothetical operation (X and Y are not in px)

% rescale X,Y extents of image (in px) to the coordinate space used to generate Z
imgx = interp1(imrange(roix),imrange(x),[1 size(inpict,2)],'linear','extrap');
imgy = interp1(imrange(roiy),imrange(y),[1 size(inpict,1)],'linear','extrap');

% draw image (must be RGB!)
if size(inpict,3) == 1
    safepict = repmat(inpict,[1 1 3]);
end
imshow(safepict,'xdata',imgx,'ydata',imgy); hold on % specify the non-px X,Y extents

% ... the X,Y data used for plotting are the same used to generate Z
[~,hc] = contourf(X,Y,roisample,10);
hc.LineStyle = 'none';

% adjust things so that dataAR allows image AR to be preserved
roiar = range(roix)/range(roiy);
contar = range(x)/range(y);
set(gca,'dataaspectratio',[contar/roiar 1 1])

colormap(parula)
colorbar
axis on % just to show the coordinate system