Corner localization and camera calibration from imaged lattices

Recovered points ( unchanged, filtered out, new corner, out of image).

This is the code for my master's thesis. The code is a bit messy, so I recommend reading the thesis first. To run the code, install pixi and run pixi shell, for example.

The code is structured as follows:

• src
  • augmentations: Augmentations for the dataset. Add an overlay image, or remove some corners.
  • benchmark: Find the error between the detected corners and the ground truth, for both real and synthetic data.
  • data: load the babelcalib dataset.
  • feature_detector: Detect corners in the image.
  • feature_refiner: Unused experimental code
  • projector: Camera model and projection functions.
  • solver
    • optimization: Optimization algorithms using Jax
    • scaramuzza: Scaramuzza's calibration algorithm, used for initialization
• main.py: Run the calibration algorithm on the dataset, entry point
• notebooks: Various visualizations, evaluation visualizations, and synthetic data generation. Also contains symbolic calculations, used in the thesis.