An integrated Empirical Mode Decomposition and Butterworth filter based vehicle trajectory reconstruction method

Trajectory data serving as an essential data-source, has been widely applied in traffic flow analysis, traffic prediction and transportation management. In real situations, trajectory data is often corrupted with noises, which may introduce estimation bias and control inefficiency to intelligent transportation systems. This paper presents a novel trajectory reconstruction method which is generic for both highway and urban arterial trajectories. The reconstruction method establishes an Empirical Mode Decomposition (EMD) based Butterworth low-pass filter framework to filter the noises and simultaneously maintain physical integrity. The two-stage framework firstly applies the EMD to decompose the original trajectories into components, multiple intrinsic mode functions (IMFs), to find out the main components of different temporal-frequency characteristics. Based on that, an optimal Butterworth-filter is applied on the lower order IMFs to filter the acceleration of an unexpected high-frequency range. To test the effectiveness of our proposed method, multiple resource data-sets are applied. As results indicated that our proposed reconstruction method performs well in terms of physical trajectories integrity, high-frequency noise removal, and measurement error rejection with minimum signal distortion. Further, our method efficiently produces speed and acceleration with higher quality compared with the state-of-the-art methods.