Unstable calibration of exterior orientation elements of dual-camera under length constraints

In recent years, dynamic stereo vision measurement technique is applying wider and wider in the field of geometric measurement, especially in wide-field measurement. A new unstable calibration algorithm is proposed under a certain amount of length constraints, which could be used in the measurement environment such as unstable cameras based on analyzing the existing stable calibration method of relative orientation for dynamic stereo vision measurement. A certain amount of calibrated length targets is distributed in the measurement area; using two cameras the center coordinates of the points on the length targets are obtained; according to the epipolar geometric constraints the dual-camera exterior orientation elements are calibrated quickly and accurately. In every measurement, the cameras are re-oriented according to the calibration targets, the calibration targets compensate the position variation of the cameras in the measurement process, which guarantees that even the positions of the cameras vary the cameras still could precisely perform the measurement. In the experiment, 90 retroreflector points distributed within a measurement range of 2.5 m×1.6 m×4.5 m were measured, the distances between 5 pairs of points were taken as the distance constraints and the results were compared with those measured using V-STARS/S. The experiment results show that the 3D reconstruction position error of the 90 points is less than 0.160 mm, the standard deviation is less than 0.042 mm when the camera positions are varied obviously. The algorithm was applied to the measurement of synthetic aperture radar (SAR) and similar result was concluded. ©, 2015, Yi Qi Yi Biao Xue Bao/Chinese Journal of Scientific Instrument. All right reserved.