A Neural Approach to Cursive Handwritten Character Recognition Using Features Extracted from Binarization Technique - Complex System Modelling and Control Through Intelligent Soft Computations - page 744

Fig. 12 Matrix representation of input sample

8 Classi cation and Recognition Process

This is the final step in an OCR System. The accuracy of the character recognition

system depends on the preprocessing techniques adopted to clean the input character
patterns. The OCR accuracy also depends on the quality of the features extracted from
the input character images to be recognized. If the correct preprocessing techniques
are applied ef

ciently and the extracted features are of good quality, the recognition
accuracy of the OCR system will be high. Apart from preprocessing technique and
the feature extraction technique, the OCR accuracy also depends on the type of
classi er involved to do the recognition. An extensive review of the literature indi-

cates that as far as the unconstrained handwritten character recognition is concerned,
neural networks as a classifier are chosen to be the best among the others.

8.1 Methodology

To accomplish the task of character classi cation and mapping, the multilayer feed

forward arti cial neural network is considered with nonlinear differentiable function

' in all processing units of output and hidden layers. The processing units in

the input layer, corresponds to the dimensionality of the input pattern, are linear.
The number of output units corresponds to the number of distinct classes in the
pattern classi

tansig '

cation. A method has been developed, so that network can be trained
to capture the mapping implicitly in the set of input-output pattern pair collected
during an experiment and simultaneously expected to modal the unknown system
from which the predictions can be made for the new or untrained set of data.