Laboratory and field tests
The feasibility of the machine vision application is typically examined by the image processing expert by means of sample parts. The perfect case would be that sample parts are available from different lots.
In the image processing laboratory, it is important to test
- whether all error characteristics are found
- which lighting and optical components are required for inspection
- which camera resolution is needed to achieve the required accuracy
- which technology is required to solve the task. Intelligent camera systems and PC systems of different designs come into question.
- which software tools are required for inspection and which cycle time can be achieved
- tests to determine precision and repeatability
The possible future variation of the samples should be included as well as the mechanical environment. The customer should also lend his experience with these important parameters.
Depending on the effort and documentation, feasibility studies can be for free or liable for costs, which can partially be set off against the order.
Tips for the laboratory work
It is compulsory to include intensive tests for interference in the laboratory, which may occur in the subsequent production process, too.
- Vibrations are easy to generate on the camera stand
- Brightness variations can be simulated by different exposure times or by changing the aperture of the lens
- Effects of extraneous light can be simulated by extra light (also only partially in certain areas of the picture)
- With some (cooking) oil, for instance, you can produce oily metal parts within seconds
- An oxidised surface is achieved by strong heating
- Dust, dirt, etc. on the transmitted light illumination are no problem at all
- Position variations of the components are also easy to test.
Create a log file of your measurements and check how they vary and how precise and repeatable they are. With these tricks and methods you are well prepared against unpleasant surprises in the real machine vision environment any you know the limits of your application.
A possibly charged field test
is performed to confirm the laboratory results. For this purpose, an image processing system is temporarily attached to the machine and connected to the PLC control in a simple way. The selected machine vision components can be tested within the production process and an overall feasibility can now be demonstrated using statistically large part quantities and all production varieties. The duration can be several days to weeks.
The results are recorded and summarised in writing. Together with the customer the results are discussed and applied to a revised specification sheet.