摘要
为解决障碍物在三维目标跟踪中产生的遮挡问题,使用基于光流法和立体视觉的目标跟踪系统,同时结合自适应滤波方法来预测跟踪目标三维速度的期望值。采用模糊控制系统解决不同速度的耦合问题,通过计算跟踪算法与预测算法的速度差值来判断运动目标是否受到遮挡。实验结果表明:在确保91.2%识别率的前提下,该算法识别运动目标平均每帧耗时15 ms。该跟踪算法具有较好的鲁棒性和实时性。
An object tracking system based on optical flow and stereo vision is used in order to solve the occlusion effect produced by obstacles in 3D object tracking, combined with adaptive filters to predict the expected 3D velocities of objects. The critical point of the system is the coupling between tracking and predictive algorithms. A fuzzy control system is proposed to solve this coupling problem between the different velocities and judging whether the moving objects are blocked by calculating the speed difference between tracking algorithm and prediction algorithm.Experimental results show this algorithm costs average 15 ms per recognition of one frame with a recognition rate at 91.2%. This tracking algorithm has better robustness and real-time ability.
出处
《中国测试》
CAS
北大核心
2015年第7期95-98,共4页
China Measurement & Test
关键词
目标跟踪
光流法
立体视觉
自适应滤波
遮挡
模糊控制
object tracking
optical flow
stereo vision
adaptive filters
occlusion
fuzzy control