Multi-Vision Tracking and Collaboration Based on Spatial Particle Filter
Long Liu, Danyang Jing · 2018
In existing multi-vision tracking methods, a distributed collaborative tracking mode based on homography constraints is often adopted, yet there are significant shortcomings to this approach. For example, visual information complementation is not used to improve the robustness of tracking, and collaborative tracking is limited by homography constraints. In this study, a three-dimensional spatial particle filter tracking method is proposed, and multi-vision joint tracking and collaboration were effectively achieved. This method is based on the existing particle filter framework. A two-dimensional plane particle is taken as the projection of a three-dimensional spatial particle on the imaging plane, and the formula for calculating a spatial particle's weight is derived based on Bayesian posterior probability recursion. In addition, an approximation method to determine spatial particle weight is given. The resampling of spatial particles is performed by using an epipolar line resampling method. The results showed that the proposed method had higher tracking precision and anti-occlusion performance than other existing methods.