FFT filtering to clean up images
ImageJ has a very easy to use FFT feature which is excellent for image filtering (and also for quantification).


Here is an example where the raw data (top images by Maria Grazia) have uneven illumination from left to right. The raw data are 512X512 pixels. By applying a high pass filter from 128 to 0 pixels, the low frequency variation is removed; the resulting bottom images are flatter.

Due to laser instability, there are also horizontal lines in the image. These may be removed by masking in the power spectrum. (Also see http://rsbweb.nih.gov/ij/docs/examples/FFT/.)

line removal

The manual drawing for filtering this one image may be turned into a universal mask by thresholding the image from 1 to 255 to use as a filter to apply to all images with this particular line issue.

This is also an extremely useful technique for images from inexpensive scanners or CCD camera electron microscopy images.