How an OCR works?

Question

I like to design an OCR for my mother toung. Can any one say from where I have to begin? Which language should I prefer?

I like to design an OCR for Malayalam (My Mother toung). Can any one suggest me that from where I have to begin? Which programing language should I prefer? From where I can find more details?

Answer 1

An Optical Character Recognition or OCR system is computer software engineered to convert hand-written or typewritten text (usually scanned) documents into machine editable text formats. Earlier versions or OCR required the system be trained to translate specific fonts, however current systems are intelligent enough to recognize most of the fonts with a high level of accuracy. Some systems also have the capability of reproducing a formatted version almost exact to that of the original page.

OCR systems uses algorithms to recognize characters, and make use of Neural Networks to increase the accuracy..

this is in simple ways, how the Neural Network wors:
-----------------------------------------------------------------
With backpropagation (an algorithm), the data is repeatedly presented to the neural network. During each of these presentations, the output of the neural network is compared to the desired output and an error is computed. This error is then fed back to the neural network and used to fine-tune the weights such that the inaccuracy decreases with each iteration and the neural model gets closer and closer to producing the desired output.

Answer 2

properly, Adobe Acrobat professional is often very stable at OCR besides. although, if the test's extremely undesirable, or if there's a hectic historic past, then function extraction could be very confusing to end.

Answer 3

I think you should read some journals about OCR. Try IEEE journals