Noise in a photographic system is generally non-additive. As of today there is no model to explain a non-additive noise system. That best model that we have of the photographic system is in terms of additive noise systems.
By adding a noise effect to a signal an additive noise effect can be obtained. What was investigated in this thesis was what effect the adding of noise in a photographic system plays in an observers ability to recognize a signal that is present in the noise.
By photographing a noise field and adding it to a signal target an additive noise effect is obtained. The method of producing the noise used in this experiment was based upon a method propogated bt N.D. Diamantides. The method used in this experiment was to disperse salt on a black background in a random manner and then photographing the resultant field.
The resultant noise fields which were normally distributed with respect to density were superposed on top of a signal slide and then shown to 25 observers. The signals used were the RIT alphanumeric targets with contrast ratios representing a useful range of pictorial photography. The observers ability to resolve a line was determined on the basis of their being able to see two out of three characters correctly.
The results of viewer observations were that the higher the noise the less the observer is able to recognize. As contrast decreases the ability to recognize a character falls off exponentially. The ability of an observer to correctly identify small lines is limited by the modulation transfer function of the eye.
Library of Congress Subject Headings
Photographic optics; Resolution (Optics)
Nastvogel, Brian, "Observer recognition of targets imbedded in noise" (1976). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus