rameau-fr/MC-Calib

What it solves

MC-Calib provides a robust way to calibrate multi-camera systems, determining the intrinsic and extrinsic parameters of multiple cameras simultaneously. It specifically addresses the challenge of calibrating hybrid systems that combine different types of cameras, such as perspective (Brown model) and fisheye (Kannala model) lenses, in a single rig.

How it works

The toolbox uses Charuco boards as calibration targets. The process involves generating these boards, capturing synchronized images of them from multiple cameras, and then running an optimization process to find the best camera parameters. It supports both homogeneous systems (all cameras of the same type) and hybrid systems (mixed types). The system can utilize a "bootstrapped" or traditional hand-eye calibration approach to refine the spatial relationship between cameras.

Who it’s for

It is designed for researchers and engineers working with multi-camera rigs, computer vision systems, and robotics where precise spatial alignment and synchronization of multiple sensors are required.

Highlights

• Hybrid Camera Support: Capable of calibrating a mix of perspective and fisheye cameras.
• Flexible Target Generation: Includes tools to generate custom Charuco boards with varying square sizes and counts.
• Robust Optimization: Uses RANSAC for outlier removal and non-linear refinement for high accuracy.
• Comprehensive Output: Generates detailed YAML files containing camera matrices, distortion vectors, and extrinsic pose matrices.