(A) study on intelligent rasterizer of chinese outline fonts and automatic generation of hinting information

Abstract

For character processing on displays and middle resolution output devices, it is the worldwide trend to represent a character with outline description. In spite of the extreme interest on the outline fonts in east Asian countries where Chinese characters are used, it is not commonly used due to the different characteristics of Chinese characters from Latin characters. The very reason lies in that it takes too much time and labor intensive jobs to design and develop a set of Chinese typeface due to the complex structured shapes and a large size of code set. The quality of small size bitmaps generated by outline fonts is more seriously deteriorated than Latin fonts below middle resolution devices. In this dissertation, we propose an intelligent rasterization scheme specially designed for Chinese characters. We also propose the automated methodology to generate the hinted outline data from the plain outline fonts for the proposed rasterization scheme. Our outline font rasterizer is focused on stem regularization. With the help of the hinting information which is added to plain outline font, the bitmap font with a good quality can be generated. We could eliminate the anomalies such as stroke disappearance, discordance of stroke width, irregular stroke intervals and stroke discontinuity problems which directly influence on the quality of Chineses characters. We present the experimental studies on the quality control methods on specific devices. One is to compensate the output difference caused by the characteristics of engine structure of 300 dpi laser beam printers. We solved this problem by suggesting another scan conversion algorithm with boundary filling. Another one is to eliminate the stroke discordance problem in representing a gray scale font on gray scale display. We applied our hinting mechanism to gray scale fonts, and show a gool result. The hinting information can be automatically generated by identifying the horizontal and vertical strokes in outline f...