Page 235 - Mechatronics for Safety, Security and Dependability in a New Era

Ch45-I044963.fm Page 219 Tuesday, August 1, 2006 3:55 PM
3:55 PM
Tuesday, August 1, 2006
Page 219
Ch45-I044963.fm
219
219
SEARCH AND POSE RECOGNITION OF
INDUSTRIAL COMPONENTS USING
CURVATURE OF OPTIMIZED EDGE PIXELS

1
Kunihiro Goto and Fumihiko Saitoh'
1
Department of Information Science, Faculty of Engineering, Gifu
University, 1-1 yanagido, Gifu-shi, Gifu, 501-1193, Japan

ABSTRUCT

We propose a method to recognize a target image area that has a free location and a free inclination in
an objective image. This method uses curvatures that are measured in two sizes of areas as the
matching key in order to improve the reliability of matching. Besides, if the only effective edge pixels
can be used for matching, the higher reliability and the low computational cost may be expected
because the number of registered template data decreases. To realize this function, the proposed
method uses the genetic algorithm (GA) to determine the optimal combination of effective edge pixels
for matching from the huge combinations of selected edge pixels.

KEYWORDS

search and pose recognition, pair of curvatures, edge pixel selection, voting process, genetic algorithm

IMAGE MATCHING METHOD BASED ON PAIR OF CURVATURES

Curvature: o) S1, o.) Li
Template table
Registration to L 8 a P D
to so oil. a 0,1 a o P» Do
tos.i 01 Pi D,
edge direction O>LA « i
l»S.i t»L.i o, « / P, D,
V J
Matching key Pose data
Figure 1: Geometrical parameters to register a template table

The proposed matching method consists of the template registration phase and the matching phase. In
the template registration phase, edge pixels are extracted in the template image. Fig. 1 shows the
geometrical relation between parameters that are stored in a template table. The variance of the edge