Introduction to Optical Character Recognition Using Neural Networks:

Optical character recognition is a technology where the software will be capable of detecting handwritten or printed matter. Human beings are able to differentiate the characters and they can identify them due to the stimulations signals which are processes by brain. We apply the same neural technology scheme in the software which can identify the characters. The software built is filled with artificial intelligence which stimulates neural networks for identifying. In this paper we will see how optical character recognition technology is developed by using neural networks and how are implemented in the software and the advantages of using this technology.

Brief on the working model:

In this we use a character recognition device which converts the characters printed on the image into machine language. Here the scanned image is converted into machine language and the output is given to the neural network. The neural networks have the capability to represent both linear and non linear systems and they can learn these relationships directly from the modeled data.

In neural networks the input data is fed to the input layer and here they are multiplied by the interconnection weights. The resultant of the input layer is passed to the first hidden layer and they get summed and it is processed by a non linear function and the obtained value is sent to the second hidden layer which is again multiplied with the interconnection weights so as to obtain the neural network output.

Conclusions:

When compared to other techniques the neural network technique is the best method for obtaining the output accurately. The neural networks have the capability to identify the imprecise and data and they derive the meaning from them. Here the quality of the output depends upon the scanned image and if the if there is improper scanned image then the quality of the output gets decreased.

Download  Electronics Seminar Report on Optical Character Recognition Using Neural Networks.