Implementation of multiple fingertip blob recognition on image processing based touch-screen for surface computers

Stating that human touch sensing surfaces have revolutionized the era of human machine interaction is perhaps stating the obvious. It is a known fact that the invention of touch-screen has almost eradicated the so-called primary input device, “the keypad” or “the keyboard” from modern electronic gadgets. The technology used in order to sense the touch is noteworthy. It is broadly categorized as Resistive touch-screen, sensing single touches and capacitive touch-screen that responds to multiple touches. The other variant is image processing based touch-screen that uses different optical solutions. Natural User Interface community has proposed numerous methodologies to implement surface computers based on imaging touch-screens. Though it is not sought-after for hand-held electronic gadgets like the other technologies because of its massiveness and input processing complexity, yet it is the only preferred technology when research in surface (tabletop) computing is taken under consideration because it extends the scope of research to object recognition too. In this paper we elucidate an algorithm for the same in order to extract multiple points of touch (fingertip blobs) on the surface, which is considered as a two dimensional plane, in the form of Cartesian coordinates. The algorithm renders accurate results with minimized construction complexity and cost of implementation. The algorithm has been implemented using MATLAB R2012a.