Online handwriting recognition using multiple pattern class models

The field of personal computing has begun to make a transition from the desk-top to handheld devices, thereby requiring input paradigms that are more suited for single hand entry than a keyboard and recent developments in online handwriting recognition allow for such input modalities. Data entry using a pen forms a natural, convenient interface. The large number of writing styles and the variability between them makes the problem of writer-independent unconstrained handwriting recognition a very challenging pattern recognition problem. The state-of-the-art in online handwriting recognition is such that it has found practical success in very constrained problems. In this thesis, a method of identifying different writing styles, referred to as lexemes, is described. Approaches for constructing both non-parametric and parametric classifiers are described that take advantage of the identified lexemes to form a more compact representation of the data, while maintaining good recognition accuracies. Experimental results are presented on different sets of unconstrained online handwritten characters and words. In addition, a method of combining information from lexeme models built on different feature sets is described, and results are presented on both English characters and Devanagari characters. Finally, a method of writer-adaptation is described which makes use of the lexemes identified from a large group of writers to define lexemes within a small amount of data from a single writer.