On techniques for the representation and synthesis of scalable Chinese calligraphic fonts
Student thesis: Doctoral Thesis
Related Research Unit(s)
Chinese computing has been expanding rapidly. To provide multiple font styles and multiple point-sizes of quality Chinese characters, the need for constructing scalable fonts is undoubtedly important. Recent researches aimed to generating quality scalable Chinese characters mainly focus on describing font shapes by smooth curves [CHAN88, ZHANG90, LIAO91, GU91]. However the smooth character outlines and solid coloring generated with this approach would not possess the aesthetic property of oriental calligraphy. We conjecture that alternative solutions can preserve the aesthetic characteristics of oriental calligraphy, and support high quality character generation. The research focus on investigating new techniques for high-quality Chinese calligraphic character generation. Three novel techniques have been developed and their results are evaluated: Fractal Outline Coding, Image Details Synthesis based on Multi-Resolution Coherence, and Brush Modeling. The first contribution of the thesis deals with a novel approach to scalable font representation based on fractal transform which we call the "Fractal Fonts". The principal idea is to automatically encode fractal font shapes from digitized shape outlines. Given the fractal characteristics of the font, characters can be regenerated in different scales with infinite details along the outlines while synthesizing pleasant looking wrinkles that simulates the effect of ink diffusion. The underling concept of the approach is the Fractal Transform Theory [BARNS93, JACQ921. While this theory has been used intensively in image encoding, it is the first time that the theory is applied to the encoding of outlines of complex shapes. We successfully formulated a fractal shape coding scheme using local Iterated Function System [MONG93, WONG931, and developed an automated shape encoding algorithm based on Fractal Transform Theory specifically for calligraphic characters. The result of this investigation have been published in [WONG94, IP94, IP961. The second contribution of the thesis concerns with the development of an image detail synthesis technique for brush-written manuscripts. The goal is to develop a technique which produces the effects of grey-level ink colors of brush writings. It should be stressed that outline font techniques are not appropriate for this purpose. Our attention then turns to image based modeling which includes the coding of minimal amount of information for each character and the extraction of calligraphy image features necessary for the regeneration of calligraphic writings. In doing so, we observe a new coherence property for calligraphic writings which we call "Multi- Resolution Coherence". This property is based on a correlation of the spatial distribution of image grey values across successive resolutions or scales. We exploit such multi-resolution spatial coherence to synthesize fine details of an image at a high resolution from a low-resolution input image which contains the basic structural contents of the image at a coarse resolution. The approach is particularly suited to synthesizing details of images containing calligraphic writings. This technique has been published in [WONG97]. The third contribution of the thesis presents a physic-based brush model. The goal is to simulate the physical process of brush stroke creation, so as to generate realistic brush written characters. Three separate physical aspects of brush writing are captured in our model, namely, the physical geometry of the writing brush, the dynamic movement, eg., speed and angle, of the brush along the stroke trajectory, and the amount of ink absorbed in the brush bundle as well as the ink depositing process. By controlling these physical parameters associated with the writing process, very realistic appearance of calligraphic writings can be generated. In particular, the aesthetic features commonly associated with calligraphy, such as the varying widths of a stroke, the impression of physical rubbing between the brush and the underlying paper, the varying shades of grey caused by different degrees of ink content in the brush, and the black and white trails created by fast movement of a drying brush can be simulated. This methodology and its results have been presented in [IP97]. In summary, we have approached the problems associated with Chinese Calligraphic Character modeling from different dimensions: fractal outline coding, image based coding by multi-resolution spatial coherence, and physical writing process modeling. While preserving the aesthetic properties of calligraphic characters is the main goal of this research, each investigated approach is also optimized to guarantee high efficiency of storage usage and short processing time for encoding and regeneration. Results obtained from the research are expected to furnish the Chinese desktop publishing industry as new alternatives to high-quality scalable Chinese calligraphic fonts.
- Data processing, Chinese characters