What would be the best way to define a device noise profile for a digital image source like a Scanning Electron Microscope? The images are scanned in a raster fashion and often there is significant noise in the resulting images. Before Neat Image I'd run these noisy images through a median filter. NI certainly offers more power and flexibility.
I'd like to build up a device noise profile for the system but obviously can't image the calibration target. What I have done (other than work up each image individually) is create a pretty good noise free image and then save the profile and apply it to subsequent images but this doesn't work quite as well across a wide variety of images of different subjects and contrasts. I've a feeling I can do better.
Any suggestions? -v
device profile for non-camera sources
-
vetenskapsman
- Posts: 1
- Joined: Tue Feb 11, 2003 6:24 am
- Location: SF Bay Area
In building profiles you need two things: image area suitable for profiling and knowledge of the device mode.
It follows from your post that you can get an image with large enough featureless areas suitable for profiling.
What you need then is to identify what parameters of your imaging device significantly influence the noise properties. For the digital cameras, there is such a list at p.12-13. For an electronic microscope, you need to identify a similar list of parameters that are significant from the viewpoint of noise.
As soon as you have this list, you can build profiles for different device modes and keep track of these modes. Then you will be able to reuse these profiles to filter other images in every device mode that you have build a profile for.
Hope this helps.
Vlad
It follows from your post that you can get an image with large enough featureless areas suitable for profiling.
What you need then is to identify what parameters of your imaging device significantly influence the noise properties. For the digital cameras, there is such a list at p.12-13. For an electronic microscope, you need to identify a similar list of parameters that are significant from the viewpoint of noise.
As soon as you have this list, you can build profiles for different device modes and keep track of these modes. Then you will be able to reuse these profiles to filter other images in every device mode that you have build a profile for.
Hope this helps.
Vlad
