The Programmable Optical 3D Measuring System POM - Applications and Performance 
Raimund Loser & Thomas Luhmann 
Leica Heerbrugg PMU 
CH-5035 Unterentfelden 
Switzerland 
ABSTRACT 
POM is a programmable 3D multi-sensor measuring system for non-contact optical measurements of points and 
elements by digital image processing and close-range photogrammetry. The system was initially developed for use in 
the automobile industry. Due to its unique open system architecture, an integrated database and program interpreter, it 
can easily be adapted to a large variety of applications. 
The system is able to work with different imaging sensors, e.g. reseau-scanning cameras, CCD cameras and video 
theodolites. A digital rotary table and a number of different light sources may be integrated. 
The control software package is based on UNIX and X-Windows/Motif. It consists of the main modules: graphical user- 
interface, SQL-based database, C-like program interpreter, 3D calculation programs and bundle adjustment, sensor 
control and image processing. 
With suitable sensor and light configurations it is possible to measure a variety of parts made of different materials, e.g. 
sheet-metal, rubber tubing, glass windshields, plastic objects etc. The powerful element-measuring algorithm enables 
the direct determination, by multi-image matching, of geometric elements such as circles, cylinders and lines in 3D 
space. Measurements can either be performed in manual mode (teach-in) or in automatic mode, where the interpreter 
processes complete measurement program files which are created by the internal program generator. 
The paper presents the current development status and describes the measurement of different objects types. Accuracy 
and speed of the system will also be discussed. 
Key Words: on-line photogrammetry, 3D vision, image processing, industrial metrology, quality assurance 
1. INTRODUCTION 
In the past few years, the technical development in 
photogrammetry, with the availibility of low-cost powerful 
computer systems, high-resolution electro-optical 
imaging sensors and image processing algorithms for 
pattern recognition and surface measurement, has led to 
the development of close-range photogrammetric 
systems for on-line and real-time applications. 
The main field of applications is in the industrial 
environment. Existing machine vision systems normally 
provide 2D image processing and analysis and are 
widely used in a large number of industrial applications. 
However, there are also a few commercial systems, 
which are more or less designed for specific sensors or 
applications and which are only used in a limited number 
of applications [e.g. El-Hakim 1986, Haggrén & Leikas 
1987, Grün 1987, Metronor 1990, AICON & VW 1991]. In 
Addition there is a large number of manual and 
automated theodolite systems which are mainly used for 
larger objects in industry [Staiger 1992]. 
This report presents the result of a joint development 
between Leica (former Kern) and Rollei in co-operation 
with several academic institutes. A first version of the 
Programmable Optical Measuring System (POM) is in 
use at Volkswagen's R&D department (Fig. 1). 
2. REQUIREMENTS 
2.1 Specifications 
The initial application at the start of development was the 
3D measurement of second-source parts which had to 
be checked for a correct fit into the car body. The 
conventional solution of this task is manual inspection 
using expensive, object-specific mechanical gauges 
[Hegelmann 1989]. The range of components that shall 
be measured by the system is characterized by very 
different features of shape, size, surface materials and 
measuring tasks. The major tasks are measurements of 
holes, edges, corners, diameters, distances and a variety 
of special object parts. These features can mainly be 
described by regular mathematical elements such as 
circles, lines and cylinders in 3D space. 
A system accuracy of +0.1mm (95%) in an object space 
of 2.0m x 2.0m x 0.6m is required (1:20'000) and has to 
be verified with distance measurements. A detailed 
investigation of accuracy is given in [Godding & Luhmann 
1992]. 
Due to internal changes of quality philosophies the 
customer has shifted the responsibility for good quality to 
the second-source manufacturers. This has led to the 
new situation in which, instead of large series 
inspections, these parts will only be checked selectively 
with respect to specific problems that might occur during 
manufacturing process. Nevertheless the system 
flexibility must be very high.