A Novel Trajectory Smoothing Algorithm Based on Empirical Mode Decomposition

This paper puts forward a novel trajectory smoothing algorithm based on empirical mode decomposition. In this method, the x and y coordinates series of the trajectories are respectively decomposed into different frequency parts through empirical mode decomposition; and then the high frequency parts of the coordinates are adaptively discarded; finally the residual and low frequency intrinsic mode functions are retained as the smoothed result. The experiments on trajectories from two different scenes show that the fluctuation in the trajectory, caused by the target deformation and occlusion, can be decreased effectively with this method and the result is better than wavelet decomposition.