Printed Text Character Analysis Version-II: Optimized optical character recognition for noisy images with the new user training and background detection mechanism

The proposed system performs the task of analysing snapshots of written text and creating fully customizable text files using Optical Character Recognition (OCR) technology. It is known that new font styles and writing formats are introduced everyday but the existing systems find it increasingly difficult to incorporate the newly emerging font styles. The authors have already proposed a system which gives the user complete liberty to effortlessly train the system to handle new fonts using the character dictionary and user training mechanism. The present version makes the process of character recognition more accurate and effective by introducing optimization in the recognition process, a mechanism to handle noisy text images and also a background detection mechanism to differentiate the written symbol from the image background.