Tuesday, September 20, 2011

HUMAN VISUAL SALIENCY
This is an intriguing topic which deals with how we humans see and understand our surroundings.  A clear working knowledge of human visual attention and fixation patterns could help researchers develop more robust algorithms to mimic human behavior more closely. In this thread, I will talk about the process of  "visual salience calculation" and existing methods for modeling them. Feel free to comment or start any related threads.

Understanding the importance of a certain image/object feature is can be modeled using the following two approaches,

  1. The top-bottom approach based on the objective of the test.
  2. The bottom-up approach which involves with the primitive sensory inputs into the human brain.
The latter is more interesting, since that models the natural human mind's selective procedure.