Application: Presence Verification
Inspection Challenge:
A manufacturer needed to inspect the labels for its products. They print ten different types of labels, and needed to know which type of label was being printed at a certain point in time. The quality of the labels (missing colors, etc.) also needed to be checked.
Needed to verify:
- Is the label the correct design? (inspect two out of every sheet).
- Are the colors correct?
Machine Vision Solution:
The following DT Vision Foundry Tools were added to the Point and Click Script to solve this inspection challenge:
- Avi Player Tool – This tool shows a variety of images for display purposes. It is ideal for rapid prototyping because it is fast – in a run-time operation, it would be replaced with the more sophisticated Picture Tool, which takes the image of the product in order for the measurements to be processed.
- Search Tool – This tool will search for the product underneath the camera and move the parameters to take into account where the product is relevant to the camera.
- Blob Analysis Tool - This tool will be used to threshold the ROI (Region Of Interest) to check the outline of the label. This tool will use its 78 different statistics to pick up on any color variations that could affect the quality of the product.
- Image Classifier Tool – This tool will have a catalog of the correct labels in memory. When the Search Tool locates a sample, the Image Classifier will tell the vision system whether or not this is the correct label.
- Text Tool – This tool visually displays the results of the test.
- Status Tool – This tool also displays the PASS/FAIL results and will also show the pass/fail rate, variables and will also save any images either passed or failed to a folder for later failure analysis to be completed.
The screen shot below shows that the image has been captured from a camera and has been displayed upon the screen in a view port. The screen first shows that the Search Tool has found the label and has relocated the ROI (Region of Interest). The Blob Analysis Tool then thresholds the ROI to achieve all the statistics needed within that ROI to calculate color variations. A color frame grabber is used.
The Image Classifier Tool will then use the same ROI to check the catalogue of images it has in its memory to tell the vision system which label is being presented underneath the camera. The result of this can be displayed on the screen.
The information is passed to the If Then tool, which will then let the system know whether the label is a Pass/Fail. As can be seen, the Text Tool displays the type of label that is underneath the camera.
The screen below shows the user that there is a very good GUI (Graphical User Interface) available within the package. This Status Tool shows the user all the vital information that is needed for the quality inspection of their product to be as efficient as possible. The statistics shown are as follows:
- Pass or Fail sign (green or red respectively).
- Pass/Fail percentages.
- Variables that the user is looking for.
- The ability to save all the statistics and pass/failed images for later failure analysis inspection.
- Warnings, e.g. in case there are five failures consecutively.
This routine will continue to check the different labels until the user/programmer is satisfied that the job is done.
From then on, there are a number of scenarios that can be implemented. The Digital I/O Tool can signal the user’s PLC to operate a robot to discard the failed articles and to let the passed articles carry on down the production line. The Digital I/O Tool can even make the production line stop if there are any faults found with the products that are being checked.
Result:
The Data Translation machine vision solution automated the inspection process, resulting in improved quality and efficiency for the manufacturer.
