Abstract

In this paper we present a low-cost system based on computer vision algorithms, which automatically inspects blow-molded plastic tubes used as outer covers of car dampers. The small size of the typical defects, the dark color and the specular reflection of this kind of black plastic parts makes the automation of this process a challenging task. Customers are becoming more exigent and they can even reject full batches of thousands of pieces only because of some non-valid samples (e.g., a limit of 10 defective tubes per million), therefore a correct quality checking is very important for companies. There are several standard industrial solutions designed for the automatic inspection of several kinds of pieces but not for these specific tubes. The proposed computer vision method obtains the tube surface aspect from the deformed laser beam projections observed by an uncalibrated camera, and identifies defects by comparing its shape with respect to the one expected. Experimental results show the suitability of the system to detect holes, burrs and deformations of these tubes, improving the quality checking process at a low cost.

BIB_text

@Article {
author = {Luis Unzueta and Javier Barandiaran and Álvaro Segura},
title = {Building a Low-Cost Device for an Automatic Inspection of Blow-Molded Plastic Tubes},
journal = {International Journal for Knowledge, Science and Technology},
pages = {81-86},
volume = {1},
keywds = {
Computer Vision, Defect Detection, Plastic Tube, Quality Inspection
}
abstract = {
In this paper we present a low-cost system based on computer vision algorithms, which automatically inspects blow-molded plastic tubes used as outer covers of car dampers. The small size of the typical defects, the dark color and the specular reflection of this kind of black plastic parts makes the automation of this process a challenging task. Customers are becoming more exigent and they can even reject full batches of thousands of pieces only because of some non-valid samples (e.g., a limit of 10 defective tubes per million), therefore a correct quality checking is very important for companies. There are several standard industrial solutions designed for the automatic inspection of several kinds of pieces but not for these specific tubes. The proposed computer vision method obtains the tube surface aspect from the deformed laser beam projections observed by an uncalibrated camera, and identifies defects by comparing its shape with respect to the one expected. Experimental results show the suitability of the system to detect holes, burrs and deformations of these tubes, improving the quality checking process at a low cost.
}
date = {2010-04-01},
year = {2010},
}