10-02-2013 10:16 AM
Hello,
I am trying to use a high speed machine vision camera (AVT guppy PRO F503C) to analyze touch screen LCD user interface modules for pixel defects, foreign particles, scratches, etc... However, this process is hampered by a dark line of pixels which tracks down the screen continously during image acquistion (both grab and continous grab). I have had a little success adjusting the camera shutter value to minimize the visibility of this line, however, this has proved to be relatively inconsistent as the shutter value seems to differ from unit to unit. I am wondering if anyone has ever run into this issue and if there is any corrective action to eliminate or minimize the effects of this line (lighting, camera settings, etc..). Here is an image:
Thanks,
Chase Fearing
Test Engineer
Subzero / Wolf
Solved! Go to Solution.
10-02-2013 11:25 AM
-hardware settings-
firewire communication
pixel format: mono 8
speed: 400mbps
Packet: 2048
-software-
imaqdx
vision development
10-02-2013 01:10 PM
You need a camera with Global Shutter.
10-02-2013 03:50 PM - edited 10-02-2013 03:51 PM
To add a little explanation to MoviJohn's response, that camera's sensor exposes a line at a time rather than all at once (a rolling shutter). This causes problems if either your lighting has flicker (like from incandescent or flourescent lights) or the part is moving while exposing.
Assuming your part is not moving, you would just need to make sure that your lighting does not flicker. If you removed your existing lighting sources and added some constant lighting (like from LEDs) then you could use a rolling shutter. Moving to a global shutter camera could help, but this could just make you get inconsistent lighting frame-to-frame if your light source is still flickering during the exposure (unless you always expose significantly longer than the flicker rate).
Eric
10-02-2013 04:07 PM
Eric,
Thanks, currently the test fixture is exposed to flourescent ambient light, but I get the same results when the lights are turned off. I've experimented with different spot lights, bar lights, etc.. but have not come across a solution. Do you think it's possible the flicker is emmitted from the LCD screen itself?
Thanks,
Chase
10-02-2013 04:20 PM
Ah, from the image I thought it was like a label but now I re-reaad your post and see that it is a screen. Generally those have backlights that flicker and the LCD itself refreshes at specific intervals. I don't have any experience imaging such a part but my guess is that if you crank up your exposure time (and reduce the aperature of your lens so it doesn't over-expose) such that it covers many refreshes then it seems like it would hide it.
Eric
10-02-2013 08:19 PM
Here is a workaround. No charge. This time 🙂
I'm assuming the line always shows up in a different position vertically in the image. If this is not the case, you may need to change the framerate so that the line moves around. The VI I have attachec take in images from your camera until the "# of Images to Consider" value is met. While it is collecting the images, it builds a sigle array of the max pixel values that came before. The resulting image will not have the line. You will need to play with the # value to figure out what works for your application.
10-03-2013 08:15 AM
MoviJOHN,
That is very generous of you. Is there any chance you could post a LabVIEW 2010 version of the VI?
Thanks,
Chase
10-03-2013 11:55 AM
I think there is a forum where you can post conversion requests. Upload it there.
10-03-2013 12:23 PM
Oh! I didn;t know I could do that! Here it is in 2010 format.