Detecting Handwriting With Inertial Sensors

As a very important input mode,Handwritten pen affect people's life.In this paper,we have studied detecting handwriting with inertial sensors,and we aim to detect handwritten pen's trajectories through the analysis of inertial signals when users are writing in the surface.This paper's goal aims to achieve a stronger robust and more accuracy of handwriting detecting system,which can easily transplant to low cost hardware.The main idea of ??this paper is first to pre-process the inertial sensor signals to remove the noise of the original signal and then detect the start-point and the end-point of the inertial sensor signal.That is to say,the plane motion part and the air motion part can be detected and respectively,using different algorithms to calculate their trajectories,the main ideas and contributions of this paper are the following:(1)According to the two inertial sensors which installed in the head and the tail of the pen,We proposed a writing model which including the translational and rotational models.to calculate the trajectory of a handwritten pen when it is written on a surface.In this way,we can reduce the cumulative error due to the quadrature integration of the acceleration.(2)We put forward a new algorithm of attitude angle error correction,we will based on attitude angle sequences which calculated by accelerometer and magnetic sensor,and attitude angle sequences which calculated by the gyroscope to misaligned alignment and opportunistic fusion.The experimental results show that the accuracy of the writing trajectory can be improved after the error correction of attitude angle.In this paper,we compared several other methods which are calculate the trajectory to our method,Study shows that the mean square error of our method is reduced by 35%.