923 
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B5. Beijing 2008 
Figure 4. Relative position of the ball-thrower, jumper and 
stereovideo capture system during the stereo-video capture 
exercise. 
2.2.2 Creating 3D colour anaglyph video clips 
Once the lineout simulation video footage was captured, it was 
downloaded into individual lineout files at a resolution of 1320 
x 1080 pixels. Each of these files was then de-interlaced using 
a free software package called VirtualDub 1.7.5. (Lee 2005). 
This produced two video files which had slightly different 
angular offsets, as if they were from the perspective view of the 
left and right human eyes. The two video files where then 
colour tinted cyan (blue and green) and red using 3D Combine 
4.2.5 (Scullion, 2008). Finally the two-colour tinted video files 
were overlaid and saved as one high resolution MPEG file with 
3D Combine. This was repeated for all lineout trials. 
2.2.3 Acquisition of Rugby Players’ Body Segment 
Movement 
2.2.3.1 Ball-throw accuracy checking device 
A device (throw accuracy measurement or TAM) was designed 
and built for this study to enable a reliable measurement of 
throw accuracy. It consisted of 28 lasers arranged in two grid 
shaped planes beaming onto 28 light dependent resistors, 
which were being analysed by a small processor for any 
incidence of a laser being broken (Figure 5). The time and 
position information was then sent to an analysis computer 
running a Matlab (MathWorks Inc, Natick, Massachusetts, USA) 
routine which calculated the parabolic flight path of the ball. 
(a) TAM Ball Trajectory Prediction 
Figure 5. Custom-built throw accuracy measurement (TAM) 
device; (a) a schematic diagram showing the projection screen, 
the device and the thrower’s position; (b) a ball-thrower 
prepares for a throw. Note the retro-targets on the thrower’s 
body. 
The computer program calculated the position where the ball’s 
flight path would have intersected the coronal plane of the 
jumper’s hands, and showed this point as 2D coordinates 
(Figure 6). Tests show that the theoretical parabolic error from 
the positioning of the lasers was 30 cm for a throw 15 m long. 
It should be noted that the position of the ball’s intersection on 
the coronal plane was predicted even if the ball did not reach 
the plane before hitting the ground. 
2.2.3.2 Body segment movement tracking 
A qualitative pilot study (Croft 2007) identified important 
parameters in determining throw accuracy: hip and torso 
orientation; elbow separation; front foot step; rear foot push; 
trunk flexion; body weight transfer; and hand-follow-through. 
Therefore, parameters describing these aspects were 
investigated in this study. In the study, all of the movement 
variables were observed to be present in either one or two of 
the throw phases, hence for the purpose of this study they will 
only be analysed in these relevant phases as in Table I. Only 
Phase I is discussed in detail in this paper. 
(b) TAM Laser Grid and Ball Tracking 
Figure 6. Prediction of the flight-path of the rugby ball as determined by the TAM device.