Recognition of Strings Using Nonstationary Markovian Models: An Application in ZIP Code Recognition

This paper presents Nonstationary Markovian Models and their application to recognition of strings of tokens, such as ZIP Codes in the US mailstream. Unlike traditional approaches where digits are simply recognized in isolation, the novelty of our approach lies in the manner in which recognitions scores along with domain speciic knowledge about the frequency distribution of various combination of digits are all integrated into one uniied model. The domain knowledge is derived from postal directory les. This data feeds into the models as n-grams statistics that are seamlessly integrated with recognition scores of digit images. A specially interesting facet of the model is its ability to excite and inhibit certain positions in the n-grams leading to Markov Random Fields. We have applied the general theory developed for processing strings of any symbols to ZIP Codes. We present the recognition accuracy achieved on a set of 20,000 ZIP Codes. The performance is superior to the present system which ignores all contextual information and simply relies on the recognition scores of the digit recognizers.