Vorschau

PDF, Englisch Download (80MB) | Nutzungsbedingungen

Abstract

Khmer inscriptions are primary sources of information on the history of Cambodia. Weathering,pollution, intentional damage or destruction and the acts of war have caused the inscriptions in a bad condition. Historians and Linguists have already intensively used the Khmer inscriptions to gain a better understanding of the ancient Khmer civilization.

Still, there has so far not been any research project using modern methods of scientific computing in order to - appropriately capture Khmer inscriptions digitally, - enhance these data using image processing techniques and transform them into 2D representations, - optimally repair or if possible supplement damaged or incomplete script, - provide tools for digital image processing and analysis of texts in ancient Khmer inscriptions, - and thereby contribute to the reconstruction, readability and preservation of these valuable historic documents.

In doing so, challenges arise that can only be addressed by the use of new concepts and methods. This is caused by the following facts: - The inscriptions usually exist as 3D data. - Their size or condition is sometimes problematic. - The number as well as the complexity of the characters is large. In regard to the possible combinations of consonants, sub-consonants and dependent vowels, they are comparatively huge.

The aim of this research project is to overcome these challenges and to contribute to the filling the above mentioned gaps. For this purpose, the original data of the inscriptions was acquired by a 3D scanning, then processed with image processing methods and transformed into a 2D representation, which could then be further analyzed. Single characters were isolated, and methods for their identification were developed. To identify the characters, their topological and geometrical structures were used. The features of the characters were captured by indices that are easy to compute. The identification of a given character was done by repeated filtering. The methods were developed by the example of the Khmer script of pre-Angkorian period but it can be extended to be used on Khmer scripts of other periods. As it does not exist Unicode for pre-Angkorian period script, a digital code was developed. Based on the designed mathematical methods and algorithms, a software tool was developed with which ancient Khmer characters and inscriptions can be processed and analyzed. This software tool is a first experimental version, which will be expanded in a further project for general use.