Markov Random Field-Based Statistical Character Structure Modeling for Handwritten Chinese Character Recognition
Issue No. 05 - May (2008 vol. 30)
This paper proposes a statistical-structural character modeling method based on Markov random fields (MRFs) for handwritten Chinese character recognition (HCCR). The stroke relationships of a Chinese character reflect its structure, which can be statistically represented by the neighborhood system and clique potentials within the MRF framework. Based on the prior knowledge of character structures, we design the neighborhood system that accounts for the most important stroke relationships. We penalize the structurally mismatched stroke relationships with MRFs using the prior clique potentials, and derive the likelihood clique potentials from Gaussian mixture models, which encode the large variations of stroke relationships statistically. In the proposed HCCR system, we use the single-site likelihood clique potentials to extract many candidate strokes from character images, and use the pairsite clique potentials to determine the best structural match between the input candidate strokes and the MRF-based character models by relaxation labeling. The experiments on the KAIST character database demonstrate that MRFs can statistically model character structures, and work well in the HCCR system.
Markov random fields, handwritten Chinese character recognition, statistical-structural character modeling
J. Zeng and Z. Liu, "Markov Random Field-Based Statistical Character Structure Modeling for Handwritten Chinese Character Recognition," in IEEE Transactions on Pattern Analysis & Machine Intelligence, vol. 30, no. , pp. 767-780, 2007.