We begin by using a small field-of-view lens and take 20 pictures of a white piece of paper. We take two exposure times (1/15 s and 1/6s) such that there are about 10,000 counts in each channel. We remove the bias (2048) from each image. We take the median of the 20 images:
We pass the above image through a median filter 20x20 pixels wide. This has the purpose of removing the high frequency components in the image.
To bring out the high frequency components, we divide the original plot by the median filtered plot, leaving only the high frequency portion.
A histogram for this image is below:
