An accelerometer based handwriting recognition of English alphabets using basic strokes

This paper focuses on an efficient method to recognize the English alphabets written in real time. For a perfect documented work or even for the formal written communication, a writer is expected to minimize the literature mistakes, i.e., spelling mistakes, grammatical errors, punctuation mark misplaced, etc. It would be interesting to device some technique if such mistakes could be recognized by non-human intervention in case of handwriting. Hence, a training-less system has been developed that is capable of recognizing the English alphabets in the real time and thereby the words, and make the suggestions to the writer regarding the wrong word. We have designed a digital pen which has an accelerometer and uses an efficient algorithm for the detection of the English letters. The technique is simple, based on the waveform analysis of the strokes made while writing a letter, the digital pen recognized the letters with the accuracy of 84% which after inclusion of the nearest mapping algorithm leads to the efficiency of over 96%. In this paper we discuss results only pertaining to recognition of capital alphabets which is not written in cursive writing style.