Research On Attitude-updating In Air Handwriting Recognition

Air handwriting recognition is a new type of handwriting recognition. Compared to plane handwriting, air handwriting breaks down the constraints of single plane, makes handwriting more freedom. On the other hand, the widely used smart devices equiped with inertial sensor provide platform for research on air handwriting, which makes it become hotspot in handwriting research field.Acceleration sensor reads is corresponding to inertial coordinates system, which always rotates along with handwriting platform when air handwriting. This makes that every accelerometer is corresponding to different coordinate system. Using these Accelerometer as raw data of air handwriting recognition makes accelerometer confused which leads to unconspicuous tendency of air handwriting and low recognition rate. In order to get high recognition rate, the attitude-updating of accelerometer is needed. This research reads accelerometer、angular velocity and magnetic flux from inertial sensor and regards them as the raw data of air handwriting, then proposes a attitude-updating algorithm based on three sub-samples of rotation vector algorithm applied to handwriting recognition.In the experiment, SVM is used to compare the recognition rate of different attitude-updating algorithm. The result turns out that the proposed attitude-updating algorithm helps improve the air handwriting recognition rate.