Main Article Content
A methodology which tracks the hand gesture depends on the integration of custom-built Micro Electro Mechanical Systems (MEMS)-based inertial sensor (measurement unit), the low-resolution imaging (i.e., vision) sensor. A 2-D gesture recognition changes in to a motion tracking gesture recognition in three dimensions. Essentially, it will show the inertial data sampled at 100 Hz and vision data sampled at 5frames/s. An Extended Kalman filter, which provides accurate human hand gesture recognition as well as tracking. The novel adaptive algorithm measures noise covariance, acceleration and angular rotational rates. The proposed method may reduce the velocity of error and also position drift by using MEMS Accelerometer sensor. To compensate for the time delay, the moving average filter used to reduce the frequency noise and then propagate the inertia of signal. A dynamic of time wrapping with DCT provides extracted feature and it gives exact value of 92.3% also individual numerical recognition with 100 ms.