Methods: A specialized target shooting paper was neatly pasted on a high-precision three-axis electronic translation stage. A linear one-way movement was set at a speed of 1 mm/s along the X, Y, and Z directions for a distance of 10 mm each. The coordinates of 3 pre-set target points were recorded at the start and end by a computer binocular vision system with a frequency of 10 FPS and stored in TXT format. The TXT files were imported to Imageware 13.0, and the straight-line lengths between the start and end were measured. The mean difference between each length and 10 mm were calculated to evaluate the correctness of the distance measurement. The linear movement and recording procedure was repeated 3 times, but the speed was changed to 5 mm/s to simulate the human mandibular movement speed. The trajectories of the 3 target points were fitted and the vertical dimensions from each track point to the fitted lines were measured. The mean difference was calculated between the vertical dimensions and 0 mm to evaluate the correctness of recording trajectories using this device.
Results: The correctness of distance measurements of the points 1, 2, and 3 were 0.06 mm, 0.16 mm, and 0.08 mm, respectively. The correctness of the trajectories of the points 1, 2, and 3 were 0.11 mm, 0.11 mm, and 0.10 mm, respectively.
Conclusion: Using this computer binocular vision device, the correctness of the recorded linear trajectories in the range of 10 mm was better than 0.20 mm.