Date of this Version
Image processing is continually hampered by the effects of noise. This paper introduces the GITR noise reduction and analysis system. It is performing removal and analysis of noise in images of a scale model of the University of Pennsylvania campus. GITR uses several algorithms including voting and median filtering to remove noise and difference of images and histogram functions for noise analysis.
Also, this work includes reports on inherent noise contained in the cameras available in the Penn GRASP Laboratory. A stuck-pixel map and analysis of any inherent noisy screen pattern is presented for four camera systems. Statistical data for all noise patterns is also given.
Future research directions concerning other noise removal algorithms and comparisons with this study, and reports of noise profiles of other cameras or camera lenses are motivated by the results of this experiment, namely study of the effect of combining using rotated or scaled data.
Scott H. Novack, "Generalized Image Translation and Realignment: The GITR Process", . May 1988.
Date Posted: 24 January 2008