Trainer Classifier & Post-Processor for Urdu OCR SDK

Abstract

Optical Character recognition (OCR) is one of the most successful applications of automatic pattern recognition. OCR technology is used to extract from filling forn1, and to automate data entry processes. It is prerequisite for speech production systems to aid the blind for reading. Here we restrict our problem to handwritten isolated characters of Urdu language to be considered for recognition. A gray scale image of the character to be recognized is obtained by scanning the character using a standard scanner having a resolution of 300 DP!. The Classifier or Recognizer for images is based upon theory of Artificial Neural Networks. The objective of proposed software system "Urdu OCR SDK" is to develop a Software Development Kit for Windows based system which can be integrated with Microsoft Visual C++ 6.0. Initially the system is designed and implemented as research project to investigate different techniques and investigate which is best and reliable for handwritten isolated Urdu characters.