Reconstruction Accuracy Analysis Underwater Three-dimensional Object-space Coordinate

Accuracy of kinematical data is of decisive importance in motion analysis particularly, the aim of this study was to assess the reconstruction accuracy in different environment and calibration volumes with different size on validation points of calibration frames with waterproof camera housings, and provide some methods for equipment exploitation in swimming.Methods: The calibration frame (2×2×1m) was placed in a swimming pool and was recorded simultaneously by two underwater synchronized cameras.The calibration of the recording space was performed using two different-sized calibration frames, the reconstruction accuracy of three-dimensional object space coordinate was determined in static condition both underwater and in the air, 4 validate markers and 12 control points were digitized in global calibrated space and local calibrated space with underwater cameras in water and air respectively.The reconstruction accuracy was determined by calculating the root mean square (RMS) error for reconstruction of four validation points on the calibration frames.Results: Direct linear transformation (DLT) methods were used to calculate reconstruction coordinate of the validate marker.The average root mean errors (RMS) in global calibrated space were: 1) 5.60mm, 10.44mm and 3.74mm underwater and 2) 4.89mm, 4.89mm and 3.96mm of air in x, y and z axis respectively;and in the local calibrated space were: 1) 5.29mm, 9.18mm and 3.02mm underwater and 2) 1.44mm, 1.11mm and 3.25mm in air.Conclusion: The results revealed that the RMS error was greater during underwater analysis, due to refraction, and it was increased particularly in the Y axis as the dimensional of the calibration volume were increased.