A Hybrid-Distance-Tree-Based Index for Large Chinese Calligraphic Characters Database

The paper proposes a hybrid-distance-tree（HD-Tree）-based high-dimensional indexing method which is to facilitate and speedup the Chinese calligraphic characters retrieval.Two steps are made in HDTree,first for every character in high-dimensional space are grouped into T clusters using hierarchy-based cluster algorithm,then the uniform start distance and centroid distance of every characters are pre-calculated and indexed by a partition-based B＋-tree.Comprehensive experiments are conducted to indicate the efficiency of our approach which is especially suitable for the retrieval for the large Chinese calligraphic characters database.