areometrics, 
i| bend) and 
uthors were 
of scoliosis. 
»asurement. 
| the project 
clinical and 
er skeletal 
idiographic 
girls is not 
only method 
ontact, non 
the three 
d offers the 
radiographic 
ere taken of 
consisted of 
single sitting 
in a frame, 
riod of time, 
ach set of 
radiographic 
nic to allow 
with X-ray 
photographs 
bility of the 
photographs 
'cond series 
progression 
ected using 
iidelines for 
2. THE MEASUREMENT SYSTEM 
2.1 Stereoscopic Camera System 
Normally two cameras are used to obtain 
stereoscopic photography. However, as only one 
measuring camera (Rollei 6006 metric camera) 
was available at the time, a system of twin mirrors 
and camera attachment was designed and 
constructed using the mirrors from a mirror 
stereoscope. The system records a double image 
on each exposure. As patient backs are somewhat 
vertically rectangular in shape the two images fitted 
quite well on to the square film format of the Rollei 
camera. 
CAMERA SYSTEM 
  
  
  
  
  
Fig. 1 Diagram of Camera System 
2.2 Camera System Calibration 
The camera system was calibrated against a test 
field in a school laboratory at a similar camera 
object distance to that which was used in the 
Clinical examination room. While there were no 
discontinuities in the calibration results, the 
281 
differences between the camera system 
determined values and actual test field values were 
not symmetrical, possibly caused by lack of 
flatness in the mirrors. To allow for this, a set of 
correction data for the model field was determined 
and recorded. The average correction was +1.4mm 
with a maximum correction of +6.3mm. Object 
coordinates of points subsequently measured with 
the system were then empirically modified 
according to the correction data before they were 
used for back shape determination. 
2.3 Patient Reference Frame 
A frame was made in an attempt to ensure the 
patient returned to a "standard" position each time 
the patient was photographed. The frame was 
made of stainless steel and attached or fixed to the 
wall of one of the examination rooms in the 
hospital. The frame contained adjustable chin 
support, shoulder pads and lower cross bar for 
hands to grip. Reference marks were placed 
around the frame to enable orientation of the 
stereo model. 
The frame and reference marks were calibrated in 
situ. A Rollei 6006 metric camera was used to 
acquire imagery (nine photographs) and image 
coordinates were measured in a Zeiss Planicomp 
C100 analytical plotter and processed using a 
bundle adjustment to deliver calibrated coordinates 
of the frame. 
2.4 Expected Accuracy of the Camera System 
The patients were about 2m from the stereoscopic 
camera system which had a nominal principal 
distance of 50mm and an effective base of 0.3m. A 
value of 0.0lmm was adopted as the image 
measuring accuracy. Simplified geometry was used 
to estimate the accuracy which could be achieved 
with the system. The initial estimates based on 
geometry were 1-2mm in planimetry (perpendicular 
to camera axis) and 3-4mm in depth (parallel to 
camera axis). 
3. THREE DIMENSIONAL RECONSTRUCTION 
3.1 Image Acquisition 
Before being photographed, the spinous processes 
on the patients backs were marked with special X- 
ray markers to show up on both normal 
photography as wel as X-rays to enable 
correlation at a later date. Details such as date and 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B5. Vienna 1996