The Introduction - The Benefits - The Background - The Problem - The Solution
|Machine Vision Clinches the Deal|
VISICON - Automatic quality control of clinched metal sheets.
Clinching is a technique for joining metal sheets. It is widely used to produce car parts, heating system components, cabinets and other structures. VISICON uses image processing techniques running on multiple PCs to inspect automatically the quality of clinched metal sheets. The VISICON quality control system reduces the time needed to check the quality of clinched joints. This allows a continuous on-line control of quality rather than the use of sampling or off-line techniques. The consequent reduction in the number of rejected boards results in substantial savings. VISICON liberates skilled workers to perform other more productive tasks. It offers an improvement in overall quality and a reduction in product variation.
The VISICON automatic quality control system reduces the time needed to check the quality of clinched joints. This allows a continuous on-line control of quality rather than the use of sampling or off-line techniques. The consequent reduction in the number of rejected boards and waste results in substantial savings. Continuous automatic control both improves overall quality and reduces variations in the final product caused by the often subjective judgements made by human operatives. The VISICON system liberates skilled workers to perform other more productive tasks.
Clinching is a technique for joining metal sheets. It is widely used to produce car parts, heating system components, cabinets and other structures. Joints are button-shaped depressions formed in sheets by using a cold press technique. Currently, quality control of clinched joints is based on manual measurements of button diameter. The effectiveness of this depends on several factors including workers' stress levels and the speed of the production line. Off-line control may lead to an unacceptable waste of material, whereas on-line control may not be feasible unless either a sampling system is adopted or the production line speed is limited. In both cases, the manual technique is unsuitable for safety-critical applications. An example of a safety-critical application involving clinching technology is in the production of boards for the frames of buildings. In this application, any fault in the joints located in the corners of the board may lead to sheet detachment with possible safety implications.
The quality assessment of a single joint shape is not a critical operation for skilled people, since tolerances are quite large. However, different materials, coatings, paintings and even reflections from the metal surface may introduce a high variability into the image of the shape. Furthermore each board has many tens of joints, some of which are more critical than others in determining the robustness of a board. The inspection of these shapes has to be made within the production time of a board which is only a few tens of seconds.
It is also important to classify different types of joint defects for diagnostic and statistical purposes. Additionally, system costs must be kept as low as possible to make any vision-based system competitive. Finally any system has to work in an industrial environment inspecting many different types of board models with different acceptance criteria and thresholds.
VISICON consists of a distributed image acquisition and processing system where a number of industrial PCs simultaneously process images of different parts of a clinched board in order to evaluate the quality of the individual joints. The PCs are managed by a central server, which controls movement of the boards along a conveyor belt and which assesses the quality of the boards based on the results from each of the PCs. The number of PCs used varies according to the board type and the speed of the production line. This enables the system to be scaled according to the requirements of the particular application
Image acquisition is performed under controlled conditions inside a dark tunnel enclosing sources of diffused light. A transform, specifically defined for the purpose, detects radially symmetric shapes and provides a concise set of parameters summarising the diameter, degree of symmetry and radial shading of the joint button. Such a transform is invariant with respect to luminance and contrast changes and is robust to noise and reflections. This calculation can be performed in real-time resulting in effective detection of joint points and classification of different types of defect.
A prototype quality-control system using two industrial PCs has been tested on some thousands of boards on a real production line. It has successfully provided an on-line, continuous control of quality recognising defective boards with a better than 95% accuracy. The prototype cost is low enough compared to that of the clinching machine to make a clear business case for its deployment.