Computers, geometry, and how they see. These are issues that will be plaguing my mind for the next five months, because of the two modules that I'm studying this semester. They sound like really exciting modules, because of numerous warnings posted by the respective lecturers.

"This is a research oriented (and a tough) module."

Don't even talk about computers yet. Do you even know how people see and recognize objects? Here is one theory...

Geons

This theory is called Recognition by Components (RBC) and it's by Irving Biederman.

Every object can be broken down into a set of primitive components, i.e. different combinations of these components represent different objects. These components are called geons. Each geon is described by a standard set of four properties called non-accidental properties:

• Edge: whether the edges are curved or straight
• Symmetry: whether the object has reflectional, rotational, or no symmetry
• Size: whether the object is constant in size, expanding, or expanding and contracting
• Axis: whether the axis of the object is straight of curved

If you codify these properties into a numerical code, every geon can be represented by a 4-digit number, with each digit representing each non-accidental property. If you use the non-accidental properties that I listed above, you can get 36 distinct geons (2 x 3 x 3 x 2).

The hypothesis is that humans break down every object they see into the individual geons to recognize them.

By breaking down the complex issue of object recognition into few simple components, we can see how this might be implemented in a computer so that a computer can recognize objects.

..................

Interesting Points of Interest - "If I could know me, I could know the univerze v(o.o)v" - Shirley Maclaine.