Calibrations

This page is up-to-date as of Parallax version 0.28.2

A calibration in Parallax is a data object that represents the intrinsic and extrinsic parameters of a stereo camera pair, plus metadata. A calibration is generated from a set of calibration points - correspondence points between object-space (3D) and stereo-image-space (4D). Once a calibration has been generated, it can be used to convert image points in the camera views to 3D object coordinates via triangulation.

Calibrations can be saved and loaded as Python Python pickle files.

The data contained within a calibration object/file includes:

• Name
• Coordinate system - this is the coordinate system of the object points in the calibration point set; it is also the coordinate system in which triangulation results are reported
• Offset - a 3D vector (in the calibration's coordinate system) which is applied to the triangulation result. Sometimes useful to compensate for systematic effects. Its default value is zero.
• Number of poses - the number of different reference frames that were given as part of the calibration set. From a probe-based calibration, this will usually be one. For a checkerboard-based calibration, it will be the number of checkerboard poses.
• Number of points - the number of calibration points per pose
• Calibration points
  • Image coordinates (for each camera)
  • Object coordinates (for each camera)
• Intrinsic parameters (for each camera)
  • Camera matrix
  • Distortion coefficients
• Extrinsic Parameters (for each camera, and each pose given - only the final pose is used for triangulation)
  • Rotation vectors
  • Translation vectors
• Reconstruction errors
  • For image-space reprojection (in pixels)
  • And for object-space triangulations (in microns)

Generating a calibration

There are several different ways to generate a camera calibration in Parallax. See the tutorials below for the most commonly used procedures:

• Calibration by Probe
• Calibration by Checkerboard