The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B5. Beijing 2008 
922 
calibration and the system test accuracy are provided in the 
results and analyses. 
Figure 2. (a) An image of the spine of a project participant and 
the imaging system. Note the round ink markers of the spine 
anthropometric landmarks in (b). 
2.2 Limb coordination in a rugby game lineout study 
The research involves four stages, namely: 1) capture 
stereovideo clips of a rugby lineout for virtual lineout display 
on the white screen; 2) create 3D colour anaglyph video clips; 
3) acquire players’ body segment movement; and 4) compute 
parameters to determine the quality of limb coordination. Two 
groups of ball throwers were recruited for the study. The 
accurate throwers were elite and seasoned players while the 
inaccurate throwers were occasional players. 
2.2.1 Capturing stereovideo clip of a rugby lineout for a 
virtual lineout scene 
A NuView, SX-2000, Stereo3D camcorder adapter (Figure 3a) 
developed by 1-0 Display Systems, Sacramento, California 
was selected for the application. The device allows recording 
of field-sequential video images of a rugby lineout throw 
(Figure 3b) onto a DV tape or onto a computer hard drive. The 
device uses a field-sequential technique to capture stereo 
images using a single-lens DV. Accordingly, the device uses 
mirrors to direct the incoming light rays onto the left and right 
viewing windows. A set of alternating shutters allows the right 
and left perspective view to be imaged onto the sensor as odd 
and even fields. Hence, each frame of video footage consists of 
a view from the left window as odd fields and a view from the 
right window as even fields. Consequently, each frame can be 
field-decoded (deinterlaced) to form left and right stereo 
images. 
(b) 
Figure 3. a) NuView 3D imaging device; b) Scene of a rugby 
lineout throw. 
To capture the 3D video footage for the virtual lineout scene, 
two standard 50 Hz DV cameras were placed perpendicular to, 
and directly in front of each real lineout in order to estimate 
where the optimal point in space was for the ball to be caught. 
Also, a Sony HDV 50 Hz camera, with NuView, SX-2000, 
Stereo3D camcorder adapter, was located directly in front of 
the lineout at a height similar to that of a throwers head (Figure 
4). The NuView 3D camera was approximately 5m behind the 
sideline, 5m behind where the thrower would stand, for optimal 
depth perception. This was determined through testing of the 
NuView hardware as demonstrated by Chong and Croft (2005). 
In the lineout simulation, thirty lineouts were performed to the 
distances close to six (front jumper), 10 (middle jumper), and 
14 (back jumper) metres as these were the distances estimated 
to be where the two, four, and six jumpers would be in a real 
lineout (Figure 4). The player formation was as follows: lifter, 
jumper, lifter, jumper, lifter, jumper, and lifter, with the jumpers 
entering the lineout at the nominated positions and the lifters 
standing at even distances between these. The jumper and 
lifters then performed each jump in their own time yet no balls 
were actually thrown, rather the players were asked to imagine 
one being thrown.