OPTICAL 3-D-MEASUREMENT SYSTEMS FOR QUALITY CONTROL IN INDUSTRY 
Dr. Carl-Thomas Schneider, Kurt Sinnreich 
AICON-Industrial Photogrammetry and Image Processing 
Rebenring 33, D-3300 Braunschweig, Germany 
PURPOSE: 
There is a wide range of applications of close range photogrammetry and image 
processing. The major fields in industry are control of production, quality 
inspection, research and development. In addition e.g. medicine or space 
technology are possible. Optical three dimensional measurement techniques have 
several advantages, e.g. short measurement time, measurement without contact, 
variable volume and location. The hardware components and the image processing 
algorithms will be described briefly and examples of industrial applications 
will be presented. 
KEY WORDS: Image processing, Machine Vision, Photogrammetry, Robot Vision, 3-D 
1. INTRODUCTION 
Qualtity control and inspection has become a 
major tool in industrial production. 100% control 
and error free production are necessary to reduce 
production costs and throw outs. The result is a 
reduced usage of resources as power or material. 
To achieve these goals new measurement and 
production techniques have to be introduced to 
allow new quality control mechanism. 
New measurement systems have to be developed to 
fulfil these control mechanisms without 
interfering the production process. Because of 
the necessity of non interfering systems those 
have to be contactless. These requirements can be 
fulfilled by optical measurement systems as best. 
The following paper will give three examples of 
those measurement systems to show the wide range 
of possible applications. The technical 
realization, advantages and disadvantages are 
described. The well known basic algorithms of 
photogrammetry and image processing are not 
described in detail. 
2. OPTICAL TUBE MEASUREMET SYSTEM 
2.1 The Task 
Nowadays tubes in an automatic production line 
are bended and fitted automatically to the 
product with robots. Unfortunatly robots do not 
always bend tubes in the nominal shape because of 
mechanical limits and other disturbances. Tubes 
that do not match with the correct shape can not 
fit into their final position on the product 
automatically. A wrong bended tube leads to an 
interruption of the production process, if the 
robots tries to fit the wrong bended tube. This 
is a very cost intensive process. Therefore it is 
necessary to detect a wrong tube shape earlier in 
the production process. The shape control of the 
tubes has to occur directly after the production 
of a tube. 
This measuring task is done usually by wooden 
gauges. These gauges have the disadvantage, that 
they do not give any numerical data about the 
tube shape but only the result ’does fit’ or 
  
'does not fit'. Additional gauges are usually 
made for only one tube shape, so for every tube 
shape a single gauge is necessary. The large 
number of different gauges that are necessary to 
control every different tube shape and the space 
in the production area they need is cost 
intensive, These disadvantages of gauges make a 
single measurement system for all tube shapes 
interesting. 
The requirements for a tube measurement system 
are a short measuring time, a contact less 
measuring and a measurement volume that covers 
all possible tube shapes. At least a sufficient 
accuracy of about * 0.5 mm has to be achieved. A 
system that fulfills these requirements based on 
photogrammetric and image processing methods has 
been developed and first practical experiences 
have been made (Schneider, 1990). 
2.2 Hardware Components 
The system consists of a measurement frame with 
fixed mounted CCD cameras (Bósemann, 1990). A 
granite plate is used as bottom platform. 
Reference points are attached to this temperature 
stable plate. The position of these reference 
points is measured with high accuracy to serve as 
control points for the camera calibration. The 
fixture of the tube is performed with an elastic 
net to avoid a deformation of the tubes due to 
their own weight (Fig.l). 
  
Fig.l: Schematic system configuration 
    
proce 
a3 e 
proce 
signa 
into 
numbe 
the 1 
has t 
tubes 
can ] 
sure 
least 
The w 
white 
and 
proje 
backg 
2,3 Sc 
The pi 
the C( 
In th: 
center 
a sec 
into s 
parts 
projec 
(Fig.2 
the  j 
points