How to distort an "image"

Hello, I have a .tiff file, with thousands of frames with data from a thermal camera, I managed to get the frames with Tiff function and I can do most of the image processing. I can display the data with imagesc and see a very reasonable result on my screen.

What I am stuck with, is on how to remove a slight distortion. I would like to avoid taking pictures of checkerboard. The cameras are stationary and detect the temps of the bead of a 3d printer. As you can see the distortion is minimal, is there a way correct this manually?

I am willing to play with the manual settings of the estimateCameraParametres I cannot figure out the process. What makes it worse is that the imshow shows an empty white canvas (vs the imagesc, which shows the attached screenshot)

Any ideas on how to proceed? I cannot even read the .tiff file with imread/show to start experimenting :(

Answer 1

[Image Analyst]

Not sure what you want to move where. But take a look at imwarp() and this link:

Spatial transformations Defining and applying custom transforms Steve on Image Processing

[Emmanouil Karamousadakis]

This screenshot is not ideal for the question you are right, but if I magnify the bottom, its a curve, not a straight line. I have other images that the curve deviates from the ideal straight line around 5 pixels in y, over 90pixels in x

Thank you though, I will check the links later later today!

[Image Analyst]

Why does it need to be "undistorted"? What's wrong with it as is?

[Emmanouil Karamousadakis]

Its need to be a straight line as it will serve as an input to another software for analysis of the thermal data . This image is just the first bead. I have 35000 of them in the tiff., where the bead rises in Zaxis.

[Image Analyst]

Not sure that's an answer. Why can't the other software take it as-is? You basicly said "it needs to be straight because it needs to be."

Did you try the link I sent you?

[Emmanouil Karamousadakis]

It can't because I have to feed this data back to the 3d model of the object that is printed. And the model is.... orthogonal :) :) I am taking the temperature and trying to relate stress analysis results with the temperature I am printing. I hope that makes sense.

yes. Transformation looks like promising. I have to figure out a function that will do what I want.

I just found out the conformal mapping, I will play a bit with geometric transformations so see what I can figure out.

Thank you for your help!