This paper discusses the use of character images to determine the parameters of an image degradation model. The acute angles in character images provide information used to find the model parameters. Three experiments are conducted to evaluate the use of characters. In the first experiment, large quantities of corners from character images are used to investigate how their contribution affects the mean and the standard deviation of the parameter estimators. In the second experiment, we focus on the relationship between the angles of the corners used in estimation and the estimation results. In the last experiment, we examine how likely the text in a common page would offer a reasonable estimation result compared to the results from experiments 1 and 2.
This document was originally published by IEEE in Seventh International Conference on Document Analysis and Recognition, 2003. Proceedings. Copyright restrictions may apply.
Available at: http://works.bepress.com/elisa_barney_smith/3/