How to select one image on montage and choose between the multiple images to select?

Hello everyone! How to select one image on montage and choose between the multiple images to select, please?It's my problem now! Please send me the code to my email please. Thanks. [e-mail]

Answer 1

[Elisa]

A little late for the person posting the question, but I had the same question and when unable to find a good answer, I wrote a small function to suit me. I'm sharing it for others to benefit. See the attached m-file.

Notes:

Create your montage adding a *single* pixel black border to each:

montageImage = montage(catImages,'Bordersize',[1 1]);

Pass this image into the function:

imgNum = selectMontage(montageImage)

The function will display the montage image in a new figure.

You will be prompted to select the subimage of interest by clicking with the mouse (ginput).

The number the subimage is in the sequence that formed the montage will be returned.

[DGM]

This doesn't work as its described. As written, it will only work for single-channel raster images. It will not work directly on a composite RGB image, nor will it work directly on the output of montage() as the synopsis suggests. It also can't work if the images are separated by any more than two rows of black pixels.

% some images
a = imread('peppers.png');
b = imread('cameraman.tif');

% montage() accepts either a cell array of images or a monolithic 4D image 
stack = {a b a b a b}; % a cell array
montageImage = montage(stack,'Bordersize',[1 1]); % montageImage is not a raster image

% since it's not a raster image, this doesn't work
imgNum = selectMontage(montageImage)

Error using sum
Invalid data type. First argument must be numeric or logical.

Error in selectMontage (line 26)
x_prof = sum(montageImage,1);
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

The output of montage() is a graphics object. Montage() is used to create and display a composite image, whereas imtile() simply creates a composite image and returns it as an output. If we already want to use montage() for display, we can just use montageImage.CData to get the underlying raster content.

As to the issues with RGB images and wider borders, these are one way of dealing with it:

% Sum rows and columns
x_prof = sum(montageImage,[1 3]); % fixes RGB issues
y_prof = sum(montageImage,[2 3]);

% Determine coordinates where borders exist
xx = find(x_prof==0);
xx = xx(1+find(diff(xx(:))>1)); % fixes wider border issues
yy = find(y_prof==0);
yy = yy(1+find(diff(yy(:))>1));

It might be best to ask where the images are coming from. If we're just dealing with arbitrarily preassembled composite images, we might have to deal with some sort of segmentation like this. Otherwise, if we are also doing the composition using montage() or simple concatenation, then we do have foreknowledge of the subimage geometry and the tiling order. From that, we can determine the image number based simply on the location within the composite image. We can also make sure that we don't produce an invalid output if we click on an empty spot in the montage (e.g. if we click the blank entry on a 2x3 montage containing fewer than six subimages)

[Elisa]

My solution may not solve all cases, and I think there was a gap as I converted what I was doing interactively into a function, but given that NOBODY had answered this question in 7 years and I didn't find anything else relevant online, it was offered to help people move forward a little. I found determining how many images wide vs tall were in the montage, and thus the element size, to be the most difficult part of solving the problem.

[DGM]

I appreciate that specific kind of motivation you mention. It's why I'm here. Had a problem, couldn't find anything in searches, wound up solving it in a days-long fit of madness, then went to find a place to testify.

I know that I often come across as a jerk, but sometimes the impression is more than intended. Apologies if that's what's happened. I suppose I have been too accustomed to barking at AI-generated slop answers these days, but being so reflexive is not fair in this case.

The utility is a start, but I want to offer ways to help you improve it, or if you consider it finished for your purposes, to help future readers make it work for them. Things like the example in the synopsis could easily lead someone into confusion. Explaining the different ways fixes could be approached seemed more illustrative than just writing my own complete answer and saying "no, use this".