目标跟踪系统中比较难解决的一个问题就是遮挡,当一个目标被另一个物体部分或全部遮挡时,跟踪的特性就会不完整或彻底消失,导致目标跟踪失败。为了解决这一问题,采用不变矩和归一化相关(NCC)对目标遮挡进行判断,然后采用目标位置和速度进行航迹外推,对大量外场视频进行仿真并把算法移植到硬件平台。结果表明,通过上述方法可以有效解决目标跟踪遇到的遮挡问题。

Occlusion is a difficult problem to be solved in the target tracking system. When a target is partially or completely occluded by another object, the tracking characteristics will be incomplete or completely disappear, leading to the failure of target tracking. In order to solve this problem, invariant moment and NCC (normalized correlation coefficient) are used to judge the target occlusion, and then the target position and velocity are used to extrapolate the track. A large number of field video were simulated and the algorithm was transplanted to hardware platform. Experiments. results show that the above methods can effectively solve the occlusion problem in target tracking.