Character Recognition through IMU Sensor and Machine Learning

Abstract

In the modern 'smart device' era, there have been attempts to device a digital pen to recognize the writing in a digital form. The successful attempts involve capturing image, applying image processing techniques, and following the application of machine learning algorithms some case. Although delivering excellent accuracy along with other benefits; this approach has several constraints such as cost, power, weight of the device, need of specific writing paper, and, most important, need for a special pen with fixed hardware. The work proposes a complete novel approach of devising digital pen tip that can be attached to any pen for serving the same purpose, however, addressing the drawbacks mentioned above and widening the scope of applicability. Advancement in digital sensor technologies and sophisticated machine learning algorithms provide key inspiration to investigate the issue from the scratch.