Camera calibration
- In French:
- Étalonnage de caméra
Definition
A procedure that determines a camera’s parameters: its internal characteristics, such as focal length, optical centre and lens distortion, and its position relative to a reference. It is carried out by photographing a known pattern, often a chessboard, from several angles. Without it, the distances and positions derived from the image are distorted.
Checked on against the references below. · Written with AI · checked against the sources Source: OpenCV, tutorial “Camera Calibration”.
Concrete examples
On the Stretch 4, Hello Robot describes a calibration procedure carried out before shipping. It provides the cameras’ intrinsic parameters (their internal characteristics) and computes the extrinsic parameters, that is the relative position and orientation, between each LiDAR and each camera, which aligns the RGB image with the depth measurements.
What it changes for you
The alignment between depth and the RGB image depends heavily on these extrinsics. Hello Robot therefore provides tools to refine them, leaving the intrinsics untouched, using RGB images and LiDAR scans of a static scene.
Additional material checked on .