This paper describes a deformable elastic matching approach to Handwritten Chinese Character Recognition （HCCR）. We assume that different handwriting variations share thesame topological structure

but may differ in shape details. The variations between differnet handwriting characters are modeled by a set of stroke displacement vectors （SDV）. According to theSDV derived

a model character is deformed gradually

in an effort to transform itself much closerto an input character. Experiments show that the proposed elastic matching model can efficientlydeal with local shape changes and variations between characters.