vision LCD cvi

[TheSpecialist]

I am new to using the vision software.  We manufacture keypads that use LCD displays (Not limited to 7 segement) and I was wondering how to detect partially illuminated or ghost images from the display?  What examples exist that can detect these defects?

[ChristopheC]

Hi,

The Vision Development Module features a golden template comparison function that might solve your problem.

Golden template comparison compares the pixel intensities of an image under inspection to a golden template. That would be a perfect LCD digit.

A pixel in an inspection image is returned as a defect if it does not match the corresponding pixel in the golden template within a specified tolerance.
Conceptually, inspection based on golden template comparison is simple: the algorithm subtracts an image of an ideal part and another image of a part under
inspection. Any visible defects on the inspected part show up as differences in intensity in the resulting defect image.

The Label Inspection example in Vision Assistant demonstrates that feature. 

https://www.ni.com/docs/en-US/bundle/ni-vision-concepts-help/page/nivisionconcepts.html

-Christophe

NI Vision R&D

[TheSpecialist]

It appears that comparing the LCD under test to a golden sample will give you a simple Pass/Fail result.  Is it possible to measure contrast ratio,intensity and backlighting?

[ChristopheC]

Hi,

The Golden Template comparison function can either return a binary containing the defects, or what might be better in your application, the grayscale image of the difference between the golden template and the image of the unit under inspection. You can extract the difference in contrast and/or intensity from this grayscale image.

The function can also threshold this image to return only defects which can be for example a segment not lit.

Hope this helps,

Best regards,

-Christophe

[TheSpecialist]

Do examples exist in CVI that perform this function?  I have a few sample pictures.  One is a normal display and the others are defects.  How would I be able to identify them?   

[Median9]

If you are using Vision Builder For Automated Inspection 3.0, i may have a script that could work for you.

Sincerely,