Fast character recognition system using expert systems
Kamalakannan Ganapathy, Chamod Fernando, A. Davari · 2005
In recent years significant research has been done in the field of character recognition by means of extracting features from the character image and recognizing by neural network classifiers. This paper deals with character recognition using expert system eliminating the neural network classification and training of data. First the character image is pre processed using thinning algorithm. Then using the thinned image, the features like horizontal line, vertical line, curved line, oblique lines, holes are extracted. Using the extracted features the rules are generated to classify the character using Java expert shell system (JESS). Since the rules are generated directly from the features and recognition of the character, the need for training of data in neural network classifier and the time consumed for this process can be avoided. Thus the character recognition system developed is faster.