PDA

View Full Version : Images from DFK 23G445 are abnorml occasionally



hehehoho
July 28, 2016, 05:58:06
Hello,

I connect three DFK23G445s to a industrial PC Advantech ARK-2121F http://www.advantech.com.cn/products/ARK-2000_SERIES_EMBEDDED_BOX_PCS/ARK-2121F/mod_EB8FD822-1CCA-4006-8AC1-DEF225EB4F59 which has three 10/100/1000Mbps Intel i210 GbE interfaces. All cameras are triggered by one Omron photoswitch. Cameras are set to 30fps to capture moving objects. I occasionally get some abnormal images as shown below:
1862
1863
1864

I really don't know what's wrong with the cameras and how to fix it.

regards,
James

Stefan Geissler
July 28, 2016, 11:55:15
Hello James

Pleas let me know the firmware version or serial numbers of the cameras. Also a correct image would be nice to compare it with the bad ones.

Thank you in advance!

hehehoho
July 29, 2016, 08:41:52
Hi Stefan

The serial numbers of the cameras are 20614836, 20614822 and 20614873. Is the firmware version necessary for you to know? I can't get it temporally.

Here is a normal image:
1866

I don't know how to describe the phenomenon in the first abnormal image with proper terminology, I just call it "RGB parsing fails". The second abnormal image is overexposure. In the third abnormal image, some pixels in a region on the left are not belong to this image.

regards,
James

Stefan Geissler
July 29, 2016, 09:23:55
Hi James

Thank you.

异常3.jpg
I must admit, I can not explain that. Looks for double exposure. Maybe the trigger signal is not clear? I do not know the Omron Photoswitch

sample2.jpg
Overexposed. It it was underexposed, too dark, I could explain Overexposed only, if the trigger signal comes at a situation, where the camera does not reset the sensor.

sample3.jpg
That could be easier: a missing data packet, but this should have no effect on the position of image data in the image.

Todo:
异常3.jpg sample3.jpg:
Enable Jumbo Frames on your network controller, if not already done.

sample2.jpg
Try another frame rate of possible, so the trigger frequency and framerate will not generate an interference.

I am aware, that I am guessing around only.

hehehoho
August 1, 2016, 03:01:25
Hi Stefan

The serial numbers of the cameras are 20614836, 20614822 and 20614873.

Here is a normal image:
1867

I don’t know how to describe the phenomenon in the first abnormal image with proper terminology, I just call it “RGB parsing fails”. The second abnormal image is overexposure. In the third abnormal image, some pixels in a region on the left are not belong to this image.

regards,
James

Stefan Geissler
August 1, 2016, 11:54:04
James

Unfortunately I do not have production data about you cameras in Europe. However, are the problems of all three cameras or of the older ones only?

hehehoho
August 5, 2016, 09:11:51
Stefan

I don't know whether the older ones have these problems. According to the camera's supplier, they seem to have no such problems. They don't know how to solve these problems too, so I have to look for help from here by myself.


regards,
James

Stefan Geissler
August 5, 2016, 09:20:26
James

I am the camera supplier in Europe. Well, not I am, I am part only.

And yes, there were problems with brightness of the cameras years ago, when the cameras ran triggered and over a long time period no trigger pulse occurred. The first images were to dark. However this does not fit into your error image. So far, my Asian colleague is right.

hehehoho
December 26, 2016, 08:13:27
OK. Thank you!