This paper presents an automated defect detection system for coated plastic components for the automotive industry. This research activity came up as an evolution of a previous study which employed a non-flat mirror to illuminate and inspect high reflective curved surfaces. According to this method, the rays emitted from a light source are conveyed on the surface under investigation by means of a suitably curved mirror. After the reflection on the surface, the light rays are collected by a CCD camera, in which the coating defects appear as shadows of various shapes and dimensions. In this paper we present an evolution of the above-mentioned method, introducing a simplified mirror set-up in order to reduce the costs and the complexity of the defect detection system. In fact, a set of plane mirrors is employed instead of the curved one. Moreover, the inspection of multiple bend radius parts is investigated. A prototype of the machine vision system has been developed in order to test this simplified method. This device is made up of a light projector, a set of plane mirrors for light rays reflection, a conveyor belt for handling components, a CCD camera and a desktop PC which performs image acquisition and processing. Like in the previous system, the defects are identified as shadows inside a high brightness image. At the end of the paper, first experimental results are presented.
[1]
Yi Lu Murphey,et al.
An intelligent real-time vision system for surface defect detection
,
2004,
ICPR 2004.
[2]
Xiang Zhang,et al.
Automatic classification of defects on the product surface in grinding and polishing
,
2006
.
[3]
K. Saito,et al.
Self-adjusting robotic painting system
,
2006,
Ind. Robot.
[4]
Yi Sun,et al.
Real-time automatic detection of weld defects in steel pipe
,
2005
.
[5]
M. D. Valle,et al.
Mirror synthesis in a mechatronic system for superficial defect detection
,
2003
.
[6]
Duke Gledhill,et al.
Surface measurement using active vision and light scattering
,
2007
.
[7]
Mutsuo Sano,et al.
A method for inspecting industrial parts surfaces based on an optics model
,
2000,
Machine Vision and Applications.