Abstract
A new multi‐stage classification scheme is proposed to recognize road/street names from handwritten Chinese postal addresses. The first stage, namely pre‐classification, is carried out by a reliable fuzzy rule‐based approach to classify an input road/street name into a corresponding cluster. The second stage, namely fine classification, is used to recognize the road/street name within a cluster by a hidden Markov model‐based approach. Based on the demand for lower error rates and higher processing speed, some fast and effective approaches within the scheme are adopted to make the scheme feasible. Moreover, several experiments show that the scheme can be used to produce very encouraging results.
Notes
Correspondence addressee