From prosopagnosia to image processing: feature enhancement module

摘要

Prosopagnosia is a term coming from Greeks words "prosop" (face) and "agnosia" (inability/impossibility to recognize or to identify it), also known as "face blind" affection. It might be present in different proportions in not so rare persons as it was previously believed. Extreme anomalies and abnormalities give important clues on how our brain is really functioning, our mind being able to reveal its secrets. "Understanding" a scene is not as simple as it seems to us even the interpretation that a little child might give to the things perceived around him. Yet, behind that little child there are months, years of knowledge acquisition. Cognitive aspects implemented in computational processes establish frames to model human actions relative to the human corresponding senses. The anatomical aspects, physiology, are modeled in simplified schemas in order to be implemented in virtual systems and machine models. Eye tracking models are examples to be considered for implementing artificial vision on robots. The paper comments, from the cognitive viewpoint, relevant recent studies on prosopagnosia, analysis the relevant features that have to be modularly treated in a face recognition system and proposes a new method of important directions detection on a face, which might constitute a new module, in addition to the existent face recognition software programs.