N <5 - 2 © OW J 
Ww 
a. U 
wv 
199 
results in coordinates in a local, but differently oriented coordinate system. Complete object measuring requires 
a scan of the object by viewing it from all directions. 
A possibility to combine the local coordinate systems is given by sensor orientation using a coordinate 
measurement machine (CMM). CMM's can use a tactile sensor in five axes up to a weight of 500 g. The lightest 
sensor unit (consists of a camera and a projector) that is obtainable today has a weight of 850 g /Breuckmann 
1994/. A metal-cutting machine tool can indeed carry weights up to 5 000 g, but they are less precise. 
While CMM's and metal-cutting machine tools represent expensive and complex carrier systems the optical- 
numerical link up method appears to be a viable alternative. 
Projector 
      
Camera 4 
Camera 1 Camera 3 
  
control point 
  
Fig. 4 Optical-numerical link-up of various object perspectives 
The optical link-up of different views of an object is subject to control points on and around the object being: 
signalised and recorded. Close range photogrammetry is a high suited method by which to record the control 
points. In the future, still video cameras will be the preferred instrument for fast and efficient coordinate 
calculations. 
An LCD-projector must be combined with 2-4 CCD sensors in a single unit, as shown in Fig. 4, for redundant 
orientation in a local coordinate system. The relationship between the different coordinate systems can be 
determined from the coordinates of the signalised points of the local coordinate system, as determined by the 
CCD sensors, and the control point coordinates of these points. The surface coordinates can be transformed 
from the local to the superior system using these 
transformation parameters. If a light sensor is required, the x 
measuring head can be reduced to a combination of an LCD- T. / | 
projector and a sensor. However the calculation of the * d À / 
coordinates of the signalised points then becomes somewhat 4 7 A A 
more complex. Firstly, the image coordinates of the control A y / d 
points must be determined in the grey level image, as in the / hs d 
redundant measuring head. In the second step the light A == 
sections of the Gray-code projection are cut with the edges of 
the elliptically mapped control points. These intersections 
allow the one-to-one allocation of phase relation and centre of Fig. 5 Correspondence of phase relation 
the ellipse so that the 3D coordinate calculation of signalised and centre of ellipse by edge 
points can be reduced to the calculation of the point of detection in grey level image and 
penetration of a straight line through a plain. Gray-code image 
IAPRS, Vol. 30, Part 5W1, ISPRS Intercommission Workshop “From Pixels to Sequences”, Zurich, March 22-24 1995