This research studies several aspects of the production documents by computers. The central focus is on the manner in which characters are represented for discrete devices such as CRT screens, laser printers, and digital phototypsetters. Image processing techniques such as digital filtering and spectral analysis are applied along with incremental scan conversion algorithms to select the pixels which are to be turned on in representing a given character. These techniques can produce both gray scale fonts for color and gray scale CRT's and binary fonts for printers and typesetters. The study attempts to ascertain which digital filters, and with that parameters, produce the most legible type. The resulting algorithms could be used in printers or composition systems for real time scan conversion of characters from outlines, permitting type in an infinite number of sizes and orientation to be produced on demand from a finite collection of outlines. A second part of the study attempts to analyze the images present in text in terms of various image processing models of the initial stages of human vision. The aim of this is to apply similar digital filters to the production of characters as is presumed to be applied by the human visual system. This, in turn may point to greater understanding of choices which should be made by font designers and by scan conversion algorithms to improve the legibility and speed of computerized output.
