In this simple example an image is convolved with a mask, thresholded,
then morphologically filtered to get a cleaner "edge detection" type
output. This is very simple but demonstrates ip in action:
Note the spreadsheet style definitions for the actions, which builds up
as we go.
This the original image
This is the convolution mask to apply to the image
This is the filtered output - scaled so zero is 128.
Here a binary image has been made by thresholding
And this mask is used to find the isolated spots
These are the isolated spots
so that they can be removed by logically exclusive ORing with the original
Changing any coefficients in the convolution mask or the threshold slider or morphology mask would cause a recalculation of any images depending on it, just like a spreadsheet. This makes testing ideas very easy. Obviously you can make complex trees of definitions, so ip allows nicer names for the definitions. "spreadsheets" can be saved and grouped as user functions.