Several approaches for the application of hidden Markov models tothe recognition of handwritten words are described. All approaches sharethe same description of words through strings of symbols. They differwith respect to the size of the vocabulary which has to be recognized.The authors distinguish between two cases: where the vocabulary is smalland constant, and where the vocabulary is limited but dynamic in thesense that it is a varying subset of an open one. The authors alsodescribe an application of hidden Markov models to the representation ofcontextual knowledge and propose some strategies to reject unreliableword interpretations, in particular when the word corresponding to theimage is not guaranteed to belong to the lexicon
展开▼