This article will show you how camcalib sensor calibration activities fit together. We will also provide links to further articles that detail each activity and show you how to get it right every time.
Sensor calibration activites
The sensor calibration process consists of two activity groups.
Intrinsic sensor calibration – characterizes properties and errors inherent to each sensor.
Extrinsic sensor calibration – measures the placement of each sensor.
We will only talk about intrinsic/extrinsic camera calibration to keep this article short and sweet.
Intrinsic camera calibration
The easiest way to understand the function of a camera is with the pinhole camera model, where every pixel has a unique ray that passes through the pinhole to see a minuscule portion of the scene. In more technical terms, each pixel corresponds to a precisely determined horizontal and vertical angle pair that allows you to perform exact angular measurements with your camera.
The only issue, most cameras are not pinhole cameras. But don't worry; with camcalib, you can easily convert almost any camera image back into a pinhole image and ignore all the complexities your lens system introduces.
More details
Extrinsic camera calibration
With the intrinsic camera calibration solved, you can combine two or more cameras to significantly up your metrology game.
Consider triangulation, an ancient technique to measure the distance of far-away objects by accurately measuring angles at known points. The first part, the precise measurement of angles, is solved with the intrinsic calibration of your camera. The second part, accurate knowledge of the cameras' points (and orientations), remains to be solved before you can simultaneously triangulate millions of voxels.
The only issue, measuring the precise locations of the cameras' focal points and viewing direction using calipers and goniometers is exceptionally challenging. You may consider fabricating your setup precisely and extracting the cameras' precise positions and orientations from your CAD model. But even then, you must contend with fabrication and mounting errors and thermal expansion and mechanical distortion of your baseline. Alternatively, use camcalib to solve this challenge with the push of a button.
Sensor calibration processes
Now that you know about intrinsic and extrinsic camera calibration, we can brief you on the various process flows you can use to calibrate your sensor application with camcalib.
In the most straightforward cases, you may be interested in calibrating only one camera's intrinsic. This may be for simple angle measurements or 6D pose estimation of AR tags for robot hand-eye calibration tasks.
If you are interested in stereo or multi-view camera calibration, your process may look like this. This is useful for depth sensing or measuring dense 3D point clouds.
With a more advanced application that includes inertial and LiDAR sensors, your calibration process will consist of additional intrinsic and extrinsic calibration steps. Applications here range from visual navigation, collision avoidance, self-localization and mapping to infrastructure cataloging.
Note, that the camcalib desktop application currently does not support LiDAR intrinsic/extrinsic calibration. The feature is currently under development.
In conclusion
You should have a good high-level overview of what camcalib does and the process sequences in place to deliver the best intrinsic and extrinsic calibration results with the click of a button.
In our next article we will show you hands-on examples of the camcalib process.