Panhinda - Offline Character Recognition System for Handwritten Articles

This paper presents an innovative technique to recognize Handwritten Articles. Proposed system is called "Panhinda". The target user group for this application would be the people who are involved with a lot of paper work on a daily basis. The proposed Character Recognition system was implemented with the capability of extracting the content of an image where the mentioned content is a hand written set of words or characters. The conversion process runs as a background process without any involvement of the user. Once the conversion is completed, User gets the capability of editing the converted text as he prefers with the aid of the Panhinda editor. This document describes the techniques for enhancing the quality of the image, character segmentation, character recognition and digital dictionaries. Noise removal, angle effects and lighting conditions are done at the pre-processing phase. After getting a quality binarized image, character segmentation will be done using Horizontal and Vertical Projection Profile method. The Support Vector Machine technique will be used to recognize the characters. Digital Dictionary will be used to capture the conflicts of the output. Error correction will be done by using a combined model of noisy channel model and natural language model. By walking through above mentioned processes handwritten article image will be converted into an editable text file. Experimenting with a set of 200 sample images, scanned through the Scanner, we have achieved a maximum recognition accuracy of 99.5% with manual error correction. Compared to existing commercial OCR systems, present recognition accuracy is worth contributing. Moreover, the developed technique is computationally efficient and consumes low memory.