Improved Particle Filtering Based Motion Target Detection Method
Yuheng Luo, Jingyun Xu, Peiliang Wang, Dongming Jiang · 2023
In the field of target tracking, particle filtering has been widely used due to its good performance, while the defects of traditional particle filtering methods in particle weight updating and initial value selection have seriously affected the tracking effect. To solve such problems, an improved particle filtering method is proposed in this paper. Firstly, the prior probability of generating initial particles is selected based on a large number of target statistics, which makes it possible to obtain better results at the early stage of tracking and improve the convergence speed and tracking accuracy; secondly, in the process of particle weight updating, a threshold partitioning combined with wavelet transform weight smoothing method is used, which effectively improves the particle degradation problem, the experimental results show that this method reduces the prediction error rate to 3.5%, which greatly improves the performance of the algorithm.