weird rasterization effect when capturing LCD screen

[david_jenkinson]

Hello,
In our test application, we are using a Basler A641f fw camera to capture images, and do pattern matches of, a phone LCD screen.  We have come across a phone that, when grabbing images, there is a walking horizontal line in successive frames, that slowly moves up the screen.  Think of when you see on TV, a camera is filming another screen in the scene.  That is similar to what I am seeing.  This has the potential to effect our pattern matches, because the location of the horizontal line may or may not be within our ROI when we do the pattern match. 

My theory is that it has something to do with the refresh rate of the LCD being close to the refresh rate of the camera ccd, both of which our out of my control. 

Does anyone know a way to compensate for this?  Has anyone else run into this phenomenon when capturing display images?


Thanks
David Jenkinson

[GuenterMueller]

Hi David.

1st idea: If the phone displays a non-changing image you can increase the exposure time to minimize the effect of the walking line.
Maybe this decreases the frame rate dramatically and is not desirable.

If this is not acceptable and you require a stable decision the following algorithm might be a starter:

If no match is found: Redo the matching process with "smaller patterns" built from about 40% of the upper/lower horizontal part of your "big pattern".
If one of these small patterns match -> Maybe it is the walking line problem and you should try again.

Hope you like it, Guenter

[McCarty]

Hello David,

LCD screens do not refresh in the same sense that CRT screens do.  LCD cells open to pass a continuous stream of light, and do not dim until instructed to produce a darker color.  I suggest you focus your troubleshooting efforts on your camera and on the driver software you are using.

If you are not using NI-IMAQdx 3.1, you may want to consider doing so.  While NI-IMAQdx is not a free download, you can download this driver software and run it in evaluation mode for 30 days.  NI-IMAQdx 3.1 is included in the Vision Acquisition Software 8.5 download available on www.ni.com/downloads.

Also check to see if you get the same walking line when filming other objects.  You may need to use a different video mode with your camera.  You can typically change the video mode in the Acquisition Parameters tab associated with your camera in the Measurement and Automation Explorer.  I suggest you experiment with different image size and frame rate settings. 

Best Regards,
T. McCarty 

[david_jenkinson]

McCarty,
This is actually the first phone we have seen exihibit this behavior, we have tested several (about 20) in the past.  Another interesting thing is that the lines only appear on the phone lcd region of the field of view, which occupies about 1/4 of the field of view.  Also, when the lcd first turns on, the line moves upward for a second, then downward really slowly, then when the display dims, it immediately moves faster, then disappears when the lcd goes blank. 
We are indeed using the latest 8.5 release, have installed all of the Q3 2007 upgrades for teststand/labview/vision.  Yes we are using IMAQdx.  Your suggestion of playing with the settings in measurement explorer is a good one, I will try that.  We are not sure yet if it will affect pattern matching, but our timeouts are such that there is enough time for the line to move out of the region of interest, it appears that way anyway.

Thanks

[AnalogKid2DigitalMan]

Out of curiosity, what happens if you rotate the phone 90 degrees with respect to the camera (or versa visa)?

[david_jenkinson]

If we rotate, the lines do indeed follow.  That is, when rototing the phone 90 degrees, the lines now move horizontally. 

So playing with NI max, manipulating the shutter speed, we can make the line thinner by raising the value, but then a thicker WHITE line appears and gets fatter.  By decreasing the value of shutter speed, the  black line gets fatter and fatter, until with really low shutter values, there is mostly black, and  a sliver of the screen walking up the screen.

So it's some very strange behavior on this particular phone, where I really don't see what if anything we can do, other than manipulating the shutter speed slightly to minimize the effect.

[McCarty]

Hello David,

You may also want to try different lighting setups.  Perhaps you could find a lighting configuration that minimizes the appearance of the line on the LCD on that phone.

Best Regards,
T. McCarty

[AnalogKid2DigitalMan]

Is this phone the same make/model as the others that do not have the artifact? Maybe it has a defect in the LCD driver or different firmware? Does it show up with the phone's internal backlight on?

Is it possible to image it in a dark location and fire a strobe or flash a led (probably a hassle to set up though) to minimize exposure time further?

Sorry for all the questions instead of answers. Sometime I solve imaging problems thinking outside the 'frame'.

-AK2DM

[GuenterMueller]

Hi David.
It seems to me that the phone is damaged.
Please correct me if I'm wrong, but I take from your words that it is not only the camera that sees the walking line. You can see it too, isn't it?
Guenter

[david_jenkinson]

Well, not really.  I cannot see the effect with the naked eye, only through the camera. 

To answer another question, this phone is different that any previous phone.

Fortunately this has not been a frequent occurence.  It does seem to be phone related, with the LCD drivers or some other issue.