CAMERA CALIBRATION FOR UNDERWATER APPLICATIONS: EFFECTS OF OBJECT POSITION ON THE 3D ACCURACY

Abstract

The purpose of this study was to investigate the effects of object position in the working volume on the accuracy of 3D reconstruction, using four different camera calibration approaches: 1) the classical DLT, 2) the nonlinear DLT, 3) the 2D plate and 4) the wand calibration. The DVideo kinematic analysis system was used for underwater data acquisition. The system consisted of two gen-locked Basler cameras (100 Hz) enclosed in housings. A dynamic rigid bar test (acquisition volume - 4.5?1?1.5 m3) was used to obtain the accuracy of the 3D reconstruction. Larger errors were found using the classical and nonlinear DLT methods. Furthermore, these approaches were affected by the rigid body position in the working volume. In conclusion, 2D plate and wand calibration methods provided more accurate results and were not affected by object position in the volume.