11-29-2012 09:29 AM
I use a NI PCIe 1433 framegrabber with a Basler spl4096-140km line camera.
Using triggered acqusition with an external source, I can get a framerate of 68500Hz in NI-MAX.
When I use the Matlab image acquisiton toolbox to acquire data, I however very often get the error message: 'ni: A frame has been dropped and the acqusition has been canceled". This happens even at much lower framerates (20kHz) and only seems to go away when acquiring data slowly.
Computer is a Dell Precision 7600 and I assume there should be no issues with RAM or the PC being slow. The error does not go away when reducing the data size (i.e. reducing the ROI to one single point that is readout).
Do you have any pointers why this error message appears and what can be done about it?
Thank you!
Reto
11-30-2012 08:05 AM
Are you just reading one line at a time? With linescan cameras, there is some overhead for each image created. If you are doing one line images, you are creating a lot of images.
Try setting the image size to 5 or 10 lines, or even bigger. Your CPU usage should drop, and the error should go away.
Bruce
11-30-2012 10:03 AM
Hi Bruce
Unfortunately, this produces the same error (i.e. acquiring 10000 line images with one trigger). Sometimes it gets through, sometimes the acquisition gets aborted due to the dropped frame. Further this way I have less control on the timing of the individual line scans.
Thanks,
Reto
11-30-2012 09:11 PM
I don't believe this is any error produced by the IMAQ driver. I suspect it is some issue in the MathWorks toolbox code that is aborting if it detects it missed processing a frame. I suggest asking MathWorks for assistance figuring out what might be going on as they maintain and support that piece of code.
Eric
03-26-2013 04:27 PM
Hello,
I encountered the same problem and I think it is because the allowcation of storage in matlab cannot catch up with that of the camera so that it will drop frames when the buffer is full.
I also want to know how to solve that.
Yang
03-27-2013 11:13 AM
Hi Yang,
I believe Blue Cheese was correct. This does not appear to be an IMAQ error. I would suggest contacting MathWorks for more information.