Handwritten Character Recognition
The project is a Handwritten Character Recognition system. It uses a pen and a pressure-sensitive pad to get input (the handwritten character) from the user and transform it into printed characters using a Neural Network. The input from the input device is stored, processed and resized before it is sent to the Neural Network. The Neural Network processes the input, recognizes what character has been written and gives an output representing this character. The Neural Network’s output is a set of bits. Different sequences of bits represent different characters. The output is processed and the printed character is produced.
For the Neural Network to recognize characters, it must be trained using different sets of input/output pairs. The more it is trained the better results it gives. Input to the Neural Network must have a specific size and pattern. A back propagation feedback for Neural Network is used . It has three layers; Input layer, Hidden Layer, Output Layer. The number of neurons in the input layer is 315. The number of output neurons in the output layer is 36. The number of neurons on the hidden layer was selected after experimenting with a different number of neurons and is 32 neurons.