Character Recognition of Handwritten Text Using Machine Learning and Image Processing

Abstract

In today’s world there have been various advancements in computing fields and as a result there is a greater need for smart devices. As soon as we come across the word ‘smart’, we immediately think of intelligence. It is intelligence and intelligent devices that have shaped modern life in the last two decades. One such example of an intelligent device is a character recognition device. Current character recognition methods work well to a good extent (98%) for typed text but this accuracy drops once the input starts becoming dynamic. This occurs when the text is handwritten or if there is a variation in its style, font, etc. We require accurate results irrespective of the dataset. Hence, we apply supervised machine learning which works by learning attributes and classifying labels. The advantage of this specific method is that it works even in the case of large datasets. In this project, we use image processing, supervised machine learning and deep learning algorithms to obtain accurate results for dynamic inputs without the loss of accuracy over a wide range of datasets.