Denoising, Contour Fitting And Texture Modeling For Calligraphic Tablet Images

Documents of ancient Chinese calligraphy are valuable cultural heritages. In this paper, the research is conducted to denoise and parameterize images of ancient Chinese calligraphy. And the main methods include: a novel algorithm for denoising the calligraphic images based on run-length statistics and the stroke characteristics of Chinese characters; a theme for generating outline font from calligraphic images; and an example based texture synthesis method for generating new calligraphic textures.First, a new method is proposed for denoising Chinese calligraphy tablet documents. The method includes two phases: a preprocessing phase in which a partial differential equation based total variation model and Otsu thresholding method are used, and a noise removal phase in which an optimal threshold selection from histogram of run-length probability density is proposed to remove some random and ant-like noises, and the improved Hough transform algorithm is used for line shape noise detection and removal.Second, we present a novel approach for generating outline font from digitized images of ancient Chinese calligraphy. Our approach consists of detecting feature points from character boundaries and approximating contour segments by parametric curves. The feature-point-detection is achieved by a statistical method based on the characteristics of a calligrapher. A database of basic strokes and some overlapping stroke components of Chinese characters extracted from the calligrapher works are constructed in advance. The relation between the noise level of stroke contours and the standard deviation of Gaussian kernel is retrieved from the database through the linear regression. Given an input character contour, the standard deviation for smoothing the noisy character contour can be calculated, and a principal component analysis method is employed to determine the feature points at the standard deviation. The feature points at a character contour are then used to subdivide the contour into segments. Each segment can be fitted by parametric curves to obtain the outline font.Third, an autoregressive model based texture generation for cursive style (Cao Shu) is proposed. The cursive style is the most free form art among styles of Chinese calligraphy. It is therefore a challenge to generate brush textures for cursive style. We collect several typical brush texture patches (BTP) to build a BTP library from the masterpiece of ancient famous calligraphers. The new BTP can be generated through the autoregressive model taking the BTP library as input. We also propose a control mechanism for the width of BTP through a stratified sampling method.