Handwriting Recognition Using Image Processing

Kevyne L Selmo

Supervised by Hantao Liu, Moderated by Alia I Abdelmoty

We've probably come across a situation where you've wrote notes on paper and typed them into word process afterwards. From a point of view, some may find this redundant. The main idea of this project is to be able to recognise handwritten characters from an image and output the results into a file(e.g. .txt). The approach that I am considering will consists of: - Pre-processing: e.g. making the image into a binary or perhaps applying some blurring to remove noise. - Segmentation: ideally we'd want to process separate each character. (may start with simple images consisting of non-joint up handwriting) Using connectivity approach to differentiate each character. - Feature Detection: obtain specific points of the character. Satisfying invariants, e.g. translation/scale/rotation. - Classification: using the points to classify what character it is based on training data. (perhaps machine learning)

Initial Plan

Final Report

Publication Form