Applications to Industrial Quality Inspection - 3D Computer Vision: Efficient Methods and Applications - page 246

Fig. 6.3 ROC curves of the inspection system, based on the measured values of ( a ) the roll angle ε ,
( b ) the yaw angle ρ ,( c ) the pitch angle λ ,( d ) all three pose angles. Solid , dashed ,and dotted lines
denote three different camera viewpoints. Note that the axis scaling in ( d ) is different from that
in ( a )-( c )
Fig. 6.4 Ignition plug inspection: Reference configuration ( left ) and three fault configurations with
the corresponding matching results, using two templates. Image scale is 0.2 mm per pixel
Krüger ( 2007 ) uses the oil cap described above as a test object to evaluate the per-
formance of three multiocular three-dimensional pose estimation methods devel-
oped in his work. The first regarded approach is the multiocular template matching
technique introduced by Krüger ( 2007 ). As a comparison, Krüger ( 2007 ) develops
and evaluates a multiocular extension of the feature pose map algorithm introduced
by Westling and Davis ( 1996 ) (Westling and Davis 1996 associate pose parame-