OBJECT RECOGNITION USING PARAMETRIC GEONs
Aswathi Hari, Sudheesh Madhavan · 2015
Abstract — In human, vision processing takes place within and between the retina and visual cortex. This involves various stages of vision processing where object recognition is an important phase. The object recognition in human means matching visual input with the structural representations of objects in brain. One of the efficient means of doing this is by using parametric GEONs, which is a subclass of basic GEONs a contribution of RBC (Recognition By Components) theory to the world of object recognition. Several neural network solutions are available to implement the concept like Kohonen’s SOM (Self Organization Map), SONG etc. A combination of basic shapes together with their relational details as a vector has to be given to a neural network which will generate different patterns for different objects, using which the objects can be recognized.