D measurement
h-resolution and
s from all around
jor orientation as
alculated by self-
cy which can be
ork configuration,
; using calibrated
ize of 2m can be
gen version
1.1 Hardware Setup
The VW system consists of three Rollei reseau-scanning
cameras (RSC) each equipped with a standard CCD
aray sensor which can be moved within the image
space of the camera (scanning principle) [Riechmann
1990]. In addition there are three wide-angle CCD
cameras for overview purposes. The RSC sensors are
mounted in a vertical plane together with a digital rotary
table, which carries a flexible part-mounting system. All
devices are fixed in a very stable steel frame. A set of
over 30 light sources is integrated in the measuring cell
(Fig. 2).
Sensors are individually controlled by a UNIX workstation
(Sun SparcStation Il) which is linked to a VME bus
containing image processing hardware and sensor
control devices.
1.2 Software Modules
The system is controlled by the Leica POM software
package which integrates a number of independent
processes and routines such as graphical user-interface,
database management, program interpreter,
photogrammetric and mathematical functions, sensor
control and image processing. The overall system control
is performed by an interpreter module which operates
with a special programming language.
The interpreter reads a number of programming
statements and is responsible for the synchronisation of
all processes. It checks and executes the commands,
calls required functions and stores information within a
network database. The programming sequences can be
generated by the user-interface, loaded from file or
entered by the operator.
The photogrammetric routines include the CAP bundle
adjustment program with self-calibration [Hinsken 1989],
spatial intersection, resection, 3D coordinate
transformation and a 3D element matching program
[Andresen & Helsch 1990]. This element matching
method is based on a grey level contour extraction in two
: Rotary Table
€ Light Sources
GI] Rollei RSC
"2 CCD Camera
or more images. In each image a list of non-
corresponding edge points is generated which can then
be matched to the 3D parameters of regular elements
such as circles, cylinders, straight lines or spheres. This
powerful algorithm is most often used for the current
application since most of the objects can be completely
described by a combination of these elements.
The input parameters of each observation (sensor
position, illumination, area-of-interest etc) and the
measurement results (image coordinates, 3D
coordinates) are stored in the database and can be
recalled at any time. This information is used to generate
an automatic measurement program for batch testing
which repeats the measurement performed manually
during teach-in. The database enables SQL
programming so that individual input and output formats
of measurement results can be created (e.g. CAD or
CAQ interfaces).
1.3 Technical Specifications
Measuring volume: 2.0m x 2.0m x 0.6m
Measured elements: points (targets), circle, cylinder,
line, sphere
approx. 30 sec per 3D element in
sequential mode
3 Rollei RSC, image format:
50mm x 50mm
3 standard CCD video cameras
Fribosa, accuracy: 5"
Measuring speed:
Metric cameras:
Overview cameras:
Rotary table:
Sigma 0 of bundle: 1.5um
RMS of object points: < 0.04mm
Distance error: < 0.1mm
1m
Controller
UNIX Workstation
Video Monitor
Fig. 2. POM hardware configuration of VW version
