A CONSTRAINED APPROACH TO MULTIFONT CHINESE CHARACTER-RECOGNITION

Recognizing multifont, multiple-size Chinese characters was a difficul t task in the area of optical character recognition (OCR). In this cor respondence, we introduce the constraint graph as a general character representation framework. Each character class is described by a const raint graph model. Sampling points on a character skeleton are taken a s nodes in the graph. Connection constraints and position constraints are taken as arcs in the graph. For patterns of the same character cla ss, this model captures both the topological invariance and the geomet rical invariance in a general and uniform way. Character recognition i s then formulated as a constraint-based optimization problem. A cooper ative relaxation matching algorithm that solves this optimization prob lem is developed. A practical OCR system able to recognize multifont, multiple-size Chinese characters with a satisfactory performance was i mplemented.