Share Email Print

Proceedings Paper

Automated crack detection on pressed panel products using image processing (Conference Presentation)
Author(s): Yeseul Kong; Hoyeon Moon; Hweekwon Jung; Gyuhae Park

Paper Abstract

Crack detection during the manufacturing process of pressed panel products is an important aspect of quality management. Tradition approaches for crack detection of those products are subjective and expensive because they are usually performed by experienced human inspectors. Therefore, the development and implementation of an automated and accurate inspection system is required for the press-forming process. In this study, we performed automated crack detection by integrating two image processing techniques with a multi-view-camera system. The first technique is based on evaluation of the edge lines which are extracted from a percolated object image. This technique could detect a crack without a reference image. Almost all of the edge lines of the panels show smooth variances of angle on the edges. When a crack occurs in panel products, an angle higher than 140 degree by the edge lines would appear, which could be used as an indication of crack presence. Another technique applies local image amplitude mapping (LAM) and compares a test image with the reference image. LAM is used to alleviate the problem associated with that the captured images during the manufacturing stage are not aligned against the reference image. The features created by LAM subtraction between the reference and test image are used to identify a crack. Before crack detection, multi-view images of a panel product are captured using multiple cameras. Afterwards, cracks are detected using both crack detection techniques based on image processing. The proposed technique is demonstrated in an actual manufacturing lines with real panel products. Experimental results clearly show that proposed technique could effectively improve the detection rate and speed for pressed panel products.

Paper Details

Date Published: 3 April 2018
Proc. SPIE 10602, Smart Structures and NDE for Industry 4.0, 106020G (3 April 2018); doi: 10.1117/12.2296490
Show Author Affiliations
Yeseul Kong, Chonnam National Univ. (Korea, Republic of)
Hoyeon Moon, Chonnam National Univ. (Korea, Republic of)
Hweekwon Jung, Chonnam National Univ. (Korea, Republic of)
Gyuhae Park, Chonnam National Univ. (Korea, Republic of)

Published in SPIE Proceedings Vol. 10602:
Smart Structures and NDE for Industry 4.0
Norbert G. Meyendorf; Dan J. Clingman, Editor(s)

© SPIE. Terms of Use
Back to Top
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?