Drift and oscillation cancelling during image series

[Marv_92]

Hi everybody,

I`m having a setup that is capturing images with a Hamamatsu C11440-22CU. However, I`m observing a oscillation of approx. 0.5 microns as well as a drift due to tensions in the sample holder and strong magnetic fields. So far I tried to use cross-correlation for each line and column of the pixel. This works okay in terms of drift cancelling but fails completely when it comes to oscillation correction. Are there better ways to digitally cancel out drift and oscillation when capturing a lot of pictures. The exposure time is approx. 0.03s.

Thanks in advance!

[wiebe@CARYA]

Better post some examples. It's hard to say anything about images without seeing them.

Best solution would be to avoid these problems, but you probably considered that.

Correcting images would ideally involve reference points. Sometimes the DUT has those reference points. If not you might be able to add them. Those reference points can be used to "calibrate" the image (or data).

In stead of correcting the image (manipulating the data), it could be beneficial to correct the measurements instead. Scaling, rotating and. Translating an image can introduce artifacts.

[Marv_92]

Hi wiebe@CARYA,

thanks for your answer.

Sure, I considered to lower the drift and oscillation by damping. Nevertheless, a correction by software seems to be required.
I do not manipulate the data. I am capturing two images with finite time difference and they are shifted by a small amplitude.

As far as I understood your suggestion, somehow I should add reference points to the image and  than check the shift of these reference points in the second image. But how do I mark reference points in an image or 2D array? Due to noise, the intensity of the pixels are changing slightly. Thus it makes it impossible to find the exact same region again. Did I understood this right? Do you have a suggestion how to solve this problem?

I attached a image that has been averaged over 30 consecutive images to visualize the displacement at the edges of the pad and the wire.