Date of Award
Master of Science
Electrical and Computer Engr
Through extensive psychophysical research it was found that the human visual system operates by transforming luminous input to the retina into a set of statistically independent signals which are later relayed to the brain. This is used to develop methods for analyzing the information content of visually degraded natural scenes to locate structure. Natural scenes are found to contain a sense of measurable order that can be differentiated from random scenes by image entropy, mutual information of different image regions, and by investigating the predictability represented by anisotropic dependencies. It is shown that heavy amounts of dust present in a scene can be separated from objects based on inherent statistical differences exploited by the described methods. These methods are carried out on a collected data set and results are presented verifying a measurable difference between distributions of dust and the variable structural content which is present in a natural scene.
Dissertation or thesis originally submitted to the local University of Memphis Electronic Theses & dissertation (ETD) Repository.
Lies, Micah John, "Information Level Analysis of Dust Degraded Scenes via LWIR Sensors" (2016). Electronic Theses and Dissertations. 1324.