基于CDPM的东巴象形文字预处理算法

Preprocessing Algorithm for Dongba Hieroglyph Based on CDPM

东巴文是一种十分原始的图画象形文字,为了表达丰富的含义,纳西先民往往在基本构字元素的基础上采用加缀或变形的方式来扩充文字,但是其中增加的额外元素给文字的特征提取及识别带来了巨大的干扰。因此,通过分析东巴象形文字的文字结构和特征,给出了基于CDPM的东巴象形文字预处理算法,该算法能够快速去除东巴字中的部分形变、离散的和具有粘连性的缀加元素,使得到的轮廓曲线能准确反映文字的本质特征。通过差异性、可扩展性、准确性和一致性等实验表明,基于CDPM的预处理算法使同类型的东巴字能够得到几乎一致的特征曲线,而不同类型的东巴字的特征曲线又能具有明显的差异性,从而为东巴文字的快速分类、检索和识别提供保证,也为其他象形文字的预处理研究提供有益参考。

Dongba hieroglyph is a very primitive pictograph. In order to express rich meanings,Naxi ancestors often use af-fixed or deformed methods to expand texts on the basis elements. However the extra elements added to it have caused great interfer-ence to the feature extraction and recognition of the Dongba hieroglyph. Therefore,the Dongba hieroglyph preprocess algorithmbased on CDPM is given by analyzing the structure and characteristics of Dongba hieroglyphics. The algorithm can quickly removepart of deformations,discrete and sticky elements on Dongba characters,so that the resulting contours can accurately reflect the es-sential characteristics of the hieroglyphics. Experiments on differences,scalability,accuracy and consistency show that the prepro-cessing algorithm enables the Dongba character of the same type to obtain almost identical characteristic curves,while the differenttypes can distinguish one from the other. This provides a guarantee for the rapid classification,retrieval and identification of Dongbacharacters,and also provides a useful reference for the preprocessing of other hieroglyphics.