Abstract: Abstract: Dongba hieroglyph is a primitive ideographic script. Many researchers have done a lot of research on the retrieval and recognition of Dongba hieroglyph, and applied various algorithms from different angles, but few of them are on the feature extraction and simplification. In view of the fact that the succinctness and completeness of character feature extraction will directly affect the accuracy and complexity of the recognition algorithm, we combine the related research of shape simplification in computer vision and present an improved algorithm which is helpful to the simplification of Dongba hieroglyph feature curve. Based on the discrete curve evolution algorithm, our algorithm further gives the critical point selection method based on the maximum area difference and the second simplification algorithm, which effectively remove the redundant points and potential anomalies in the character feature curve. The universality and robustness experiments show that our algorithm can remove more than 87% redundant points in the curve while retaining the original character features, and achieving the most simplified feature curve. It lays the foundation for retrieval and recognition of Dongba hieroglyph.

Key words: Keywords: extracting feature of Dongba hieroglyphs, feature curve simplification, discrete curve evolution algorithm, secondary simplification algorithm