MATLAB: [Image processing] How to match similiar but misplaced objects

image processingImage Processing Toolbox

Hi! I've got a problem with processing CT images – I want to make difference images, but my original and contrasted image are misplaced (about few pixels), so my difference image isn't as good as it could be. I've been wondering about segmentation my whole object (e.g. inverting segmentation of background), but what to do next? I work on grayscale images.

Best Answer

Have you considered imregister() to align the images. If alignment is not what you want to do (not your final objective) then explain further, because "segmentation of whole object" is very vague. Needing to align things so they can be subtracted optimally is a very different thing than segmenting your image - like classifying some or all of the parts such as liver, lungs, spine, spleen, etc.

Related Solutions

MATLAB: Image segmentation and blobs

The best one is the one that works for your images. Obviously there is no one single algorithm that works best for all natural images. Well, maybe not so obvious, because you asked. Eventually all images come down to blobs. You do whatever you have to do to get down to a gray level image that you can threshold into foreground and background blobs.

MATLAB: [Image Processing] Remove a part of a figure

If you can be sure that the top one is also left most, then just label and use ismember(). Something like

[labeledImage, numberOfRegions] = bwlabel(outlineImage);
outputImage = ismember(labeledImage, 2:numberOfRegions);

If it's above but not necessarily to the left, you could transpose (to flip it on its side), then do the above code, and transpose back.

Best Answer

Related Solutions

MATLAB: Image segmentation and blobs

MATLAB: [Image Processing] Remove a part of a figure

Related Question