融合感知哈希特征的时空上下文目标跟踪算法

摘要

文中提出融合感知哈希特征的时空上下文快速目标跟踪算法,在光照剧烈变化和目标被遮挡情况下能准确地跟踪目标.通过感知哈希特征对目标的位置进行评估,避免时空上下文算法在光照剧烈变化的跟踪漂移问题,并通过更新目标哈希值和时空上下文信息来进行下一次迭代.实验结果表明结合目标时空上下文信息的感知哈希跟踪算法与目前主流跟踪算法相比表现出更快的跟踪速度,并且在复杂背景和被遮挡后表现出较好的鲁棒性.%A real-time and fast algorithm that combined spatio-temporal context and perceptual Hash is proposed in this paper.The proposed tracker can track the target accurately and robustly when there are illumination and shelter.The spatio-temporal context was used to estimate the target's position to avoid tracking drift.Then it updates the Hash and spatio-temporal context a computing the target position in next frame.The experimental results indicate that compared with currently algorithms the algorithm combined spatio-temporal context and perceptual Hash algorithm has faster calculating speed.