Comparison of Film and Video Techniques for Three-Dimensional DLT Repredictions

Abstract

The precision of the kinematic values depends upon the methods of recording a subject’s motion. With the introduction of video recording techniques, questions have arisen concerning the accuracy of video compared with that of 16-mm film. Accordingly, the purpose of this study was to compare the accuracy of the two techniques for point reprediction using the Direct Linear Transformation method. Range poles, serving as boundaries of a cube with 20 known spatial coordinates, were filmed and videotaped. The 20 control points on the film and video recordings were digitized by three individuals. Nine sets of digitized points (three digitizers × three trials) for both film and video were compared with the actual three-dimensional coordinate values. Resultant mean errors were statistically significantly different (p<.05), 4.8 mm and 5.8 mm for film and video, respectively. However, from a practical standpoint the video error was only .29% of the calibrated field compared to .24% for film. Thus it is concluded that video techniques are comparable in accuracy to 16-mm filming methods.