Graphics Reference
In-Depth Information
Chapter 6
Applications to Industrial Quality Inspection
Industrial quality inspection is an important application domain for three-dimens-
ional computer vision methods. Traditional vision-based industrial quality inspec-
tion systems primarily rely on two-dimensional detection and pose estimation al-
gorithms, e.g. relying on the detection of point and line features, the extraction
of blob features from binarised images, or two-dimensional grey value correlation
techniques (Demant, 1999 ). More advanced vision-based quality inspection systems
employ three-dimensional methods in order to detect production faults more reliably
and robustly. In this section we regard applications in the automobile industry of the
methods for three-dimensional pose estimation of rigid and articulated objects de-
scribed in Chap. 2 .
A typical area of interest is checking for completeness of a set of small parts
attached to a large workpiece, such as plugs, cables, screws, and covers mounted
on a car engine. A different task is the inspection of the position and orientation
of parts, e.g. checking if they are correctly mounted but also checking for proper
grasping and transport with an industrial robot.
Section 6.1 regards the three-dimensional pose estimation of rigid parts in the
context of quality inspection of car engine components. The approach of object
detection by pose estimation without a priori knowledge about the object pose is
analysed in Sect. 6.1.1 , while the technique of pose refinement based on an ap-
propriate initial pose is regarded in Sect. 6.1.2 . Here we adopt the presentation of
von Bank et al. ( 2003 ) and Barrois and Wöhler ( 2007 ), respectively (cf. Sects. 2.1
and 5.6 ). The three-dimensional pose estimation of tubes and cables in the sce-
nario of car engine production is analysed in Sect. 6.2 according to d'Angelo et al.
( 2004 ) (cf. Sect. 2.2.1.2 ). For each scenario we compare our evaluation results to re-
sults reported in the literature for systems performing similar inspection tasks. Sec-
tion 6.3 describes applications of the integrated approaches introduced in Sects. 5.1 -
5.4 (Wöhler and Hafezi, 2005 ; d'Angelo and Wöhler, 2005a , 2005b , 2005c , 2006 ,
2008 ; Wöhler and d'Angelo, 2009 ; Herbort et al., 2011 ) to the three-dimensional
reconstruction of rough metallic surfaces.
 
Search WWH ::




Custom Search