Capture properties of interactive elements on a executing program

In graphical user interface testing (GUI Testing) for desktop-based/website, all of record and replay tools can capture interactive elements on a executing program. In Ranorex (http://ift.tt/2fxQOoC) or Selenium (http://ift.tt/zDsxQY), they can capture most of elements which you clicked/selected/etc.

For example, you start your program, then you click a button "OK". As a result, Ranorex/Selenium detect the exact type of this button, its content ("OK"), its location on the screen, etc.

How these tools can capture these elements? Any predictions based on your knowledge; or any scientific papers on this question are appricated.