A Bayesian framework for deformable pattern recognition with application to handwritten character recognition

Deformable models have recently been proposed for many pattern recognition applications due to their ability to handle large shape variations. These p roposed approaches represent patterns or shapes as deformable models, which deform themselves to match with the input image, and subsequently feed the extracted information into a classifier. The three components-modeling, ma tching, and classification-are often treated as independent tasks, in this paper, we study how to integrate deformable models into a Bayesian framewor k as a unified approach for modeling, matching, and classifying shapes. Han dwritten character recognition serves as a testbed for evaluating the appro ach. With the use of our system, recognition is invariant to affine transfo rmation as well as other handwriting variations. In addition, no preprocess ing or manual setting of hyperparameters (e.g., regularization parameter an d character width) is required. Besides, issues on the incorporation of con straints on model flexibility detection of subparts, and speed-up are inves tigated. Using a model set with only 23 prototypes without any discriminati ve training, we can achieve an accuracy of 94.7 percent with no rejection o n a subset (11,791 images by 100 writers) of handwritten digits from the NI ST SD-1 dataset.