Image PostProcessing FAQ

Why do the images look “muddy”?
The contrast levels are set initially to make images contain all of the available data in the image, but often this means that they seem to have a high background and structures are not well distinguished.  In ImageJ or Photoshop or GNU Image Processing package (GIMP), simply adjust the brightness and contrast to highlight the feature and structure of interest. This is best done after the image is cropped, scaled, rotated, and placed alongside other images to make a publication plate or figure.

Why do the images look noisy (salt and pepper)?
With serial block-face EM image sets, there is a trade off between signal-to-noise, cut thickness and resolution in x,y. When the project requires highest x,y resolution (approx 4.5nm/pixel) and ideally z-resolution (1.8kV=~25nm) as well, a trade off is usually required involving suboptimal scan speed to reduce the damage to the block-face (see Skipping).  First attempts with each sample set the scan scan speed for least noise and the desired resolutions and cut thickness, and then operators reduce scan speed as needed until there is no skipping.   There are other strategies to reduce beam damage effects – deeper cuts, reduced x,y resolution and different kV settings (ie z-resolution).

How can noisy images be fixed?
We offer “Enhanced Registered images” have some noise reduction applied to the full sized image after the images are registered (aligned). We usually use ROF-denoising and some CLAHE – local contrast enhancement.  All can also be done from ImageJ/FIJI.