|
|
Fuel Gauge
This application has
been taken from the book Industrial Image Processing by Christian Demant, Bernd
Streicher-Abel and Peter Waszkewitz, published by Springer (Berlin, Heidelberg,
New York), ISBN 3-540-66410-6.
The image shows a
section of a dash board with some printing defects. The "r" of the "Ltr" mark is
missing and the petrol pump exhibits a white spot in its display area.
 Typically,
the basis of a print quality inspection is a difference image, containing the
differences between the current test piece and a reference image. The creation
of a difference image for print quality inspection or any other type of
reference image comparison is far from trivial. Because of positioning and image
capturing inaccuracies, edges are practically never aligned precisely so that a
simple subtraction would lead to the detection of severe pseudo errors along the
edges. This is illustrated by the adjacent image. The pronounced errors are due
to a misalignment of a perfectly correct print.
To deal with this
problem, NeuroCheck's Print Quality Inspection function uses a highly
configurable, sophisticated three-stage positioning process and also takes heed
of the structure size of the objects to be checked. The parameters can
accommodate any desired distinction between tolerable and inadmissible defects.
Furthermore, the type of error (missing print, additional ink, both) can be
selected also. The image shows faultless detection of all defects by a
hierarchical template matching algorithm.
Due to the consistent
object-oriented approach of NeuroCheck you are not restricted to a single more
or less flexible function when build print quality inspection applications. The
difference image can of course be analyzed with the full range of functions in
NeuroCheck.
|
 |
