PDA

View Full Version : Unexpected problems with DFK 23UM021



QuaziRandom
January 11, 2014, 13:01:13
Hi,

We recently bought a DFK 23UM021 (USB 3.0 CMOS) camera from an Indian dealership (Menzel Robovision). We successfully received the product and in the process of testing it for our application, we stumbled upon a few problems. The camera doesn't seem to work well where there are fast moving subjects in the working plane although the subjects are illuminated with ample lighting. It has been taken care that the subjects are moving slow enough to be properly captured considering one of camera's shutter speeds (1/5000s). But, even with all the measures taken care of, the camera is capturing images which exhibit some kind of an after-image (not motion blur) of the moving subjects. Two samples of these are attached (wish I could attach more than 3 files totally to explain it better). An image with static subject is also attached (Static.jpg) which shows the absence of such after-image in static subjects. Secondly, it has also been observed that with faster shutter speeds, the captured images seem to be getting color filtered (purple/pink) and thus, distorting the image information. It would be wonderful if one of the community members could help us out in sorting out these problems.

FYI, our setup is illuminating the subject/s (Nuts falling down a chute) with 10W LED bar lights (cool white color) with a working distance of 100 mm. The camera is configured @ 1280x960 resolution with a frame rate of ~60 FPS. The lens used is a varifocal (2-8mm focal length) Fujinon industrial lens (Extensive tests were done to find if the problems we are facing were due to lens, and it has been found not to be so). The latest USB drivers were downloaded from the Imaging source website (usbcam_2.6.5). To capture the images, both IC Capture 2.2 and IC Imaging Control Demo Application from the latest SDK (Version 3.3) were used and found to be producing identical results. Also, it was tried on both Windows 7 And Windows 8 running PCs.

Thanks in advance.

Aravind Rao / QuaziRandom

Stefan Geissler
January 13, 2014, 11:38:21
Dear Aravind Rao,

I suppose, the problem is caused by the M021 sensor. It seems to be kind of blooming.

What are the differences in the camera settings in the image of the static nut and the moving ones?

QuaziRandom
January 14, 2014, 04:05:05
Dear Aravind Rao,

I suppose, the problem is caused by the M021 sensor. It seems to be kind of blooming.

What are the differences in the camera settings in the image of the static nut and the moving ones?

For the static image, the exposure is 1/1337 s with a gain of 32 (0dB) and pictures with the moving subjects are of 1/2500s @ gain of 32 and 1/5000s @ gain of ~45. The gain numbers are as shown in the IC Capture 2.2 software.

If it's a problem with the sensor, are there any solutions to avoid it? And, are you referring to the sensor on this particular camera we have received (which might be defective) or for the family of camera?

Also, is there any way I might be able to upload images of full resolution in this forum, apart from the option of hosting it on a public domain? I ask this because the uploaded images are heavily compressed and seem to have lost the details of the "after-image/blooming".

Stefan Geissler
January 14, 2014, 08:16:30
I spoke again with our optical engineers: It is a problem of the combination of sensor and lens. If the using a matching lens, the green pink colors should be away. Unfortunately I have to check, what a suitable lens is.

QuaziRandom
January 14, 2014, 16:10:12
I spoke again with our optical engineers: It is a problem of the combination of sensor and lens. If the using a matching lens, the green pink colors should be away. Unfortunately I have to check, what a suitable lens is.

I would like to point out that it is likely not the problem of the lens, as quite intuitively, the lens wouldn't play any role in any kind of dynamic image distortions. It would be wonderful if you could explain how does an "unmatched" lens affect this magnitude of blooming? I might be wrong after all.

Please check with your engineers and let me know if you might stumble upon any other solution.

Stefan Geissler
January 14, 2014, 16:20:19
I would like to point out that it is likely not the problem of the lens, as quite intuitively, the lens wouldn't play any role in any kind of dynamic image distortions. It would be wonderful if you could explain how does an "unmatched" lens affect this magnitude of blooming? I might be wrong after all.

Unfortunately this is wrong, because we tested this and got different results with different lenses. Some showed this green pink effect and otherones did not.

QuaziRandom
January 14, 2014, 16:32:13
Unfortunately this is wrong, because we tested this and got different results with different lenses. Some showed this green pink effect and otherones did not.

Oh, well. OK. Thanks.

Is there a catalog, from where we can pick a matching lens for our application? And just to re-assure, is it guaranteed, if chosen the right lens, that this blooming will be absent?

Stefan Geissler
January 15, 2014, 11:25:52
Is there a catalog, from where we can pick a matching lens for our application? And just to re-assure, is it guaranteed, if chosen the right lens, that this blooming will be absent?
Unfortunately, the answer to your questions is "no". Sorry for that. I am working on this.

Stefan Geissler
January 16, 2014, 10:34:58
Hello

after discussing with my colleagues I must admit, I have no solution for your problem. Therefore, you must decide, whether you want to continue testing the camera or give it back to your camera dealer.