Abstract:
Character recognition is the translation of handwritten or typewritten text into
machine-editable form. Character recognition systems can contribute
tremendously to the advancement of the automation process and can improve
the interaction between man and machine in many applications, including
office automation, preserving old/historical documents in electronic format and
a large variety of banking, business and data entry applications. The task of
isolated character recognition is a pattern classification problem, where an
unknown input pattern is to be assigned to one out of a number of given
classes. The main theme of this thesis is to prove that combined horizontal
and vertical binary histograms could be used to recognize character using
Neural Network. Histogram values of scanned normalized characters are
treated as features for recognition. Computing histogram is a much simpler
process as compared to other feature extraction methods; the proposed
approach, therefore, needs less computational time as compared to Statistical
methods, Structural Recognitions methods (strings and graphs matching) and
template Matching.
