We are wanting to capture information from windows applications and web pages in a dynamic, configurable way.
This should include being able to OCRing portions of the window, matching average pixel colors, single pixel checks, comparing a region with a sample or template image.
All of these techniques and more should be configured and tested via a user interface. The user interface should allow an image to be loaded into it (or window selected, which will have an image take of it) and regions marked out, along with any of these checks and rules on what data it generates.
For example, a rectangle would be drawn and positioned on the sample image, then an average pixel color method would be selected and a color selected, a key/name is then given for this, along with a rule that it is a boolean (another example of data type would be string) and it's TRUE if the average color matches.