Model-based stroke extraction and matching for handwritten Chinese character recognition

This paper proposes a model-based structural matching method for handwritte n Chinese character recognition (HCCR). This method is able to obtain relia ble stroke correspondence and enable structural interpretation. In the mode l base, the reference character of each category is described in an attribu ted relational graph (ARG). The input character is described with feature p oints and line segments. The strokes and inter-stroke relations of input ch aracter are not determined until being matched with a reference character, The structural matching is accomplished in two stages: candidate stroke ext raction and consistent matching. All candidate input strokes to match the r eference strokes are extracted by line following and then the consistent ma tching is achieved by heuristic search. Some structural postprocessing oper ations are applied to improve the stroke correspondence. Recognition experi ments were implemented on an image database collected in KAIST, and promisi ng results have been achieved. (C) 2001 Pattern Recognition Society. Publis hed by Elsevier Science Ltd. All rights reserved.