A hybrid approach of hidden Markov model and fuzzy logic for isolated handwritten characters recognition

Research in off-line handwriting recognition has been prevalent for many decades. After many years of intense research, it still remains an open problem. The challenging nature of handwritten characters and words recognition has attracted the attention of researchers from industry and academic circles. The commercial sector has shown significant interest in handwriting recognition research due to the large number of applications that exist. In recent years, techniques for recognizing handwriting have become more sophisticated in dealing with real-world situation and to increase recognition rates. This thesis reviews all aspects of handwriting recognition research, from the recognition of handwritten numerals to handwritten words. The different methods employed by researchers are mentioned and the approaches adopted for the research are elaborated. The focus of this thesis would be the recognition of isolated handwritten characters, concentrating on the slightly more challenging group, lowercase English alphabets. The main aim of this research work is to present a hybrid approach of a syntactical method with a statistical method in a manner that will require less training data but still yield reasonable recognition rate and high reliability rate. The system will be designed with the use of Hidden Markov Model (HMM) as a linguistic variable quantifier for a Fuzzy rule based classifier. This hybrid method, as far as according to the result of the literature search is concerned, is among the first in the area of handwriting recognition. The main advantage of this approach is a less training intensive method that does not rely on data abundance to achieve a good recognition result. The whole system that integrates the two approaches is tested against a standard database. A favorable outcome of the recognition results, has proven the approach is comparable to many other approaches as in the literature.

Preview PDF FSKTM 2011 15R.pdf Download (839kB) | Preview

Actions (login required)

View Item