Abstract:
Identification cards(ID cards) are considered to be the most reliable source for extracting citizen’s/visitor’s information.Some organizations like educational institutions, housing societies , offices , use their visitor’s ID cards to save their visitor’s information.The registration process, in Pakistan, is still done by using the traditional method of data entry i.e the manual entry that is done by writing down data on a form and then it is later on converted to digital data by retyping the information.Therefore,to make the traditional and hectic method of data entery obsolete. We’ve developed "Smart Shanakhti Card Reader" a character recognition system for Pakistani ID cards.The extracted text are then validated it and later saves it to a database for future use.This system recognizes Urdu text from Urdu CNICs and English Text from Smart ID cards respectively.We have used the deep learning techniques like Convolutional Neural Network for the recognition of Urdu text while the English and numeral text can be recognized by using existing models of Tesseract,respectively.We have used .Net Framework(C# language) for designing the interface and the classification of ID cards ,pre-processing, model training and postprocessing is done by using Python Language. The text recognition system is developed by using following steps: text area extraction, pre-processing ,text recognition and postprocessing.
