Print Registration Quality Control Using LabVIEW and IMAQ

Author(s):
Dr. J Stewart - Expert Monitoring Ltd.

Industry:
Consumer Goods

Products:
Vision, LabVIEW

The Challenge:
Monitoring a production line of printed metal used in the manufacture of aerosol cans.

The Solution:
Developing an automated visual inspection system using IMAQ products controlled by LabVIEW.

Introduction
Expert Monitoring Limited (U.K.) successfully installed a high-speed machine vision system that accurately detects, measures, and rejects print defects in the manufacture of aerosol containers. The system integrates LabVIEW and a National Instruments PCI-1424 digital image acquisition board with a high-speed line-scan camera to capture and process 160 images/minute of objects 3.00 by 5.00 mm traveling at 2.00 m/s to an accuracy of 0.05 mm.

System Design
We designed the system to find and measure the first and last registration marks of every printed metal sheet as it exits the printing press and reject any sheets that fail. The registration mark is a 3.00 by 5.00 mm cross-hair used by the printers to check that each colour is printed exactly on top of the previous colour. The press prints the sheets at varying rates of around 80 sheets per minute, and the sheets pass the camera at a typical speed of 2.00 m/s.

Two optical sensors connected to digital I/O lines on the PCI-1424 image acquisition board trigger the 1024 pixel Dalsa Spider line-scan camera. The triggers detect the front and rear edges of the sheet at the moment the first and last registration marks pass the camera. When triggered, the camera then acquires a 1,024 by 256 pixel image at a line rate of 36 kHz. To make reliable measurements, we must calculate the resolution of each image (typically 20 pixels/mm) by measuring the speed of every sheet as it passes between the triggers.
Using standard National Instruments IMAQ Vision programs (VIs) and sophisticated custom-built algorithms, all created in LabVIEW, we use the known geometry of the printed design to find the center of the registration mark and subsequently measure its dimensions and any misregistration. We make the pass/fail decision by comparing the registration error in the X and Y dimensions to user-set thresholds.
The system rejects misregistered sheets at 4.50 m past the camera station. Another optical sensor, connected to the third digital I/O line on the PCI-1424, detects the sheets as they reach the reject bin. The software counts the number of sheets passing the reject sensor, and therefore tracks passed and failed sheets along the line. We use the fourth digital I/O line as an output signal to the reject bin, telling it if the next sheet should be passed or rejected.

Challenges
Because of the movement of the sheets and the lack of a constant reference (each print design differs considerably), we had to find the registration mark using a complex series of measurements. Because we had to achieve all image processing - including acquiring the image, noise reduction, intensity amplification, finding the registration mark, and measuring - within 200 ms, we tested several methods to become increasingly more efficient by finding the smallest possible regions of interest before applying intensive processing.
Scratches and ink marks are common on the printed material. We had to set sensitivity levels carefully to ignore scratches but pick up faint colours, such as pink and grey. Furthermore, we had to modify the normal methods of detecting edges and lines so the geometry of the registration mark is not affected by miscellaneous blobs.

Because the speed of the print line is not constant, the period between the image acquisition and the sheet rejection also varies. Reading the reject sensor, counting the sheets in/out, making the pass/fail decision, and supplying the reject signal required careful timing to ensure there was no interference with the triggering of the image capture, which is software controlled.

Results
With the camera system, the client can improve customer quality assurance by ensuring that misregistered sheets do not continue through production to become aerosol cans. One sheet makes 30 cans that would be extremely difficult to remove at a later point. The system has had a further beneficial effect because problems with the machine set-up are highlighted immediately, and we can rectify them during the process. We can also investigate the source of print errors, thereby reducing scrap sheets in the long term.

For more information, contact:

Expert Monitoring Limited

Cardiff Business Technology Centre

Senghennydd Road, Cardiff CF2 44AY U.K.

Tel +44 (0)29 20372311

Fax +44 (0)29 20373436

E-mail sales@expertmon.com

Web: www.expertmon.com

Browse All Case Studies »