Calibration and autocalibration of a stereo vision system

Version du 8 janvier 2008 à 16:51 par Sdrouin (discuter | contributions)

(diff) ← Version précédente | Voir la version courante (diff) | Version suivante → (diff)

Projects > Calibration and autocalibration of a stereo vision system

Calibration consists in calculating the intrinsic and extrinsic parameters of a stereo vision system using a known calibration pattern. The autocalibration problem consists in obtaining these parameters without using of precisely known pattern. The autocalibratio methods of are advantageous if the system must be calibrated while it's being used or when a suitable calibration pattern is not available.

The first part of this report describes the model and the parameters representing a stereo vision system; for the cameras, the pinhole model is used. The fundamental matrix is used to represent the epipolar geometry.

The second part of this report presents a calibration algorithm and an autocalibration algorithm using the fundamental matrix. For calibration, an initial estimate of the intrinsic and extrinsic parameters is obtained using the relation between a plan and its image. A bundle adjustment is then used to refine the parameters of the system and the calibration pattern. For autocalibration, the intrinsic parameters of each camera are initially estimated using the Kruppa equations, then the extrinsic parameters of the system are evaluated by a decomposition in singular values of the essential matrix.

More information: Postscript, PDF, Matlab.

This work was carried out in the Summer of 2000 at LVSN with the support of a NSERC undergraduate student research award.