Fryer, John 
  
  
  
  
  
robbery. Fortunately some structural Tra itidha — ehs L 
features usually remain, often those and Camera 
cupboards, desks and other furniture used Cal brat on 1 
by bank staff. It is those features which the Spac Rese iom |t 
photogrammetrist must use in any 
  
reconstruction. A | HF T ll se 
S. Grile 1 
The forensic photogrammetrists must be 
aware that the present day bank scene will 
probably have been altered from the time of me 
hysi al c 
  
  
  
  
the incident. It is important to contact the 
manager and ask, and record in the field 
  
  
  
  
  
  
book, details of what changes have been Measurements | 
made as a defense attorney will certainly - 
pose those questions in court. It is most 
important to check if the present Y 
  
  
surveillance camera is the same one which [—3À————»| 
was there at the time of the robbery and if 
its position has ever been moved. Scientific 
evidence can be discredited easily if any 
flaw, or even if the hint of uncertainty, can 
be raised in court. 
  
  
Measure 
image 
Coordi atm 4v 
à N 
ME A cli 
Hm 
; rs 
Figure 2. Traditional Photogrammetric C | 
4 oto he ghs t 
Technique for Height Determination. by photo. 
tec nidu s e 
  
  
  
  
  
  
  
  
  
  
  
4 OBJECT SPACE RECONSTRUCTION 
Given the likely time delays between an incident and the involvement of the forensic photogrammetrist, it is strongly 
suggested that the photogrammetrist think and act in a more innovative manner than merely follow the traditional 
resection/intersection procedure outlined above. The time honoured technique outlined above (Figure 2) may prove to 
be the only possible solution to a given set of circumstances. Working in image space, however, when many 
uncertainties of camera calibration are present, really only provides ‘ammunition’ for the defense attorney when cross 
examination of the evidence takes place in court. 
The concept of object space reconstruction outlined here is based on actual experiences in court room situations. The 
physical reconstruction of a single image is a concept which the jury of lay persons, and the judge, can understand and 
embrace. Figure 1 showed the extension of the ray of light which has come from the camera, across the top of the head 
of the robber, and onto a permanent fixture such as the teller's grille. 
If a piece of string is held tautly between the grille and the camera, the actual ray of light at the time of the robbery can 
be physically recreated. This is most compelling evidence for a jury. Lens distortion and camera calibration are totally 
irrelevant to this ray of light. Even if a fish-eye lens were installed, this principle would hold true - the ray of light 
which just grazes the head of the robber as seen on the image could be faithfully reproduced in object space. 
The location of the position of the robber's feet in object space is the final consideration. Experience has indicated that 
this is probably achievable within 1 or 2 cm if a patterned carpet is present, otherwise approximately 10 - 15 cm can be 
expected if the robber is reasonably close (within a metre say) to an upright fixture such as the wall in front of the teller. 
The next consideration of a photogrammetrist is the slope of the string. The author's experiences have been of the order 
of 1:3 or 1:4, meaning that a 3 or 4 cm error in the position of the surveying staff held at the feet location will result in a 
height error of 1 cm. See Figures 3, 4 and 5 for examples of the method applied in practice to the image shown as 
  
  
248 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Amsterdam 2000. 
Fi 
gr