"A Software System for Image Processing Chinese Character Fonts"
Fonts used in computerised type-setting are typically stored as libraries of
bit-maps corresponding to individual characters. These maps are constructed by
digitizing hand-drawn masters and the data is then stored in a format suitable
for efficient use by the typesetting output device. In the case of Chinese
character fonts, the volume of data involved means that as much automation of
this process as possible is desirable.
Limitations of the character digitization process introduce noise features into
the bit-maps, decreasing the accuracy of reproduction of the characters. Two
such features commonly encountered are aliasing of near-horizontal and
near-vertical lines, and small local errors such as cavitation of edges.
This thesis describes a modular set of procedures developed to detect and
eliminate these undesirable features in the character bit-maps of various fonts
and convert this enhanced data to an edge-based vector format, suitable for
modern typesetting devices.
Damian Conway /