基于UPF-BP神经网络的视频跟踪研究

本文提出一种UPF粒子滤波与BP神经网络相结合方法用来跟踪运动背景下的视频多运动目标，该方法可以处理尺寸变换问题，从而可以鲁棒地跟踪目标。这种算法可以把最新的观测考虑进去，跟踪目标既可以是刚性目标，也可以是非刚性目标，并且跟踪算法可以实时实现。

The video tracking based on UPF and BP neural network

In this paper,UPF of a particle filter and BP neural networks methods used to track the movement against the background of the video and more moving targets,ways to deal with the changing size of the problem,which can be robust to track the target.This algorithm can take into account the latest observation,the goal is to track the target can be rigid,or non-rigid targets,tracking and real-time algorithm can achieve.