System combines motion control and machine vision

Oct. 1, 1999
Industrial-automation applications frequently combine the need for computerized motion control and machine vision. Such applications often also require automated inspection of product labels to verify accuracy.

System combines motion control and machine vision

Industrial-automation applications frequently combine the need for computerized motion control and machine vision. Such applications often also require automated inspection of product labels to verify accuracy.

"Stringent FDA regulations require the highest accuracy in labeling of pharmaceutical and medical products," says Steve McCool, president of Graftek Imaging (Austin, TX). Because of this, medical-product companies are replacing tedious and error-prone human inspection methods with automated systems.

In a semiautomated application developed for a pharmaceutical vendor, product labeling needed to be inspected off-line after each change of product type on an automated processing line. "In this application, products are produced in physically connected sets of from two to eight units," says McCool.

Because of the resolution required to identify label content of these products, it was cheaper to move each product into the camera field of view for inspection. To verify the color of the top part of the product label, Graftek used a 768 x 494 NTSC and Y/C TK-1370 color camera from JVC to inspect half of the label and a 12-in. 768 x 494 l XC-75 monochrome camera from Sony Electronics (Park Ridge, NJ) to inspect the other half.

Using the StillColor software option from National Instruments (Austin, TX), color images from NTSC or PAL color cameras were acquired with the IMAQ PCI-1408 image-acquisition board. StillColor acquires color images using multiple frames from an NTSC or PAL cameras. With StillColor, three successive frames are acquired to produce a color image. The color image acquired from an NTSC or PAL camera has color quality comparable to RGB camera and is best suited for color applications where the object is stationary because the algorithm uses multiple frames to create the color image.

Installed in a Pentium 166 with 32 Mbytes of RAM from Ampro (San Jose, CA), running Windows/NT, the PCI-1408 is used for the image-acquisition board while the National Instruments AT-MIO-16E-10 data-acquisition board is used for position sensing and motion control.

In operation, the product fixture is placed in position. Then a computer-generated work order that specifies the contents and the text present on product labels is scanned. This work order contains either a printed product description or barcoded information. Once the LabLRead system recognizes the content of the work order, it inspects the product to compare product labeling with the specified content. Both label text content and color regions are inspected to assure accuracy.

Using LabVIEW for motion control, the product fixture is moved laterally until an individual product is under the cameras. This process is repeated until the entire product batch has been inspected. After completion of the inspection process, a test report label that contains product run identification information, pass or fail information, and the reason for failure if any is then printed from LabVIEW.

Voice Your Opinion

To join the conversation, and become an exclusive member of Vision Systems Design, create an account today!