Camera properties

[mathewpeters]

Hello.
Is it possible to do this task without the camera properties(projection matrix) which is in the dino_ps.mat file. Or can you suggest me some reference or code which does this task without camera properties as I'm given with the same task of 3d reconstruction with multiple images but no camera properties are given

[zinsmatt]

Hi,

You will need the camera poses to use this method. If you don't have them, you can have a look at more advanced methods called Structure-from-Motion (SfM). They can estimate both the scene structure and the camera relative poses.

Here is an example of textured reconstructed mesh I could obtain with Kitware's open-source SfM pipeline (KWIVER+TeleSculptor): https://blog.kitware.com/3d-reconstruction-from-smartphone-videos/

Otherwise, COLMAP is also a very good alternative: https://colmap.github.io/tutorial.html

[mathewpeters]

Okay .
Can you please provide me some code snippet which generates a ".mat" file and stores the camera parameters of every image that I have.
it would be really helpful.
Thank you

[zinsmatt]

I did not generate this ".mat" file myself as it was provided in the Dino dataset, but I guess the scipy.io.savemat function can do it.
You might have to change the first lines of my code depending on how you structure your projections matrices, but, in the end, the projection variable should contain a list of 3x4 matrices.