摘要
针对目标跟踪过程中出现的定位偏差问题,提出了Mean shift和轨迹预测相结合的运动目标跟踪算法。该算法首先根据目标已知位置信息采用最小二乘法拟合运动轨迹并得到预测位置,然后利用Mean shift算法得到目标最终位置。通过计算搜索误差判断是否发生严重遮挡情况,并给出相应处理策略。进行了一系列实验,验证了算法的有效性,并将实验结果与其他算法比较,表明该算法有效地提高了快速运动目标跟踪的精度,具有较强的鲁棒性。
Aiming at the problem of location bias in the tracking process,a moving target tracking algorithm based on Mean shift and path prediction is presented.According to known position information,least squares method is used to fit a path and predict the next position of the target.Then Mean shift algorithm is utilized to obtain the final target position.To determine whether there is a serious occlusion,search error is calculated.Corresponding treatment strategies are given.A series of experiments are conducted to verify the effectiveness of the algorithm.By comparing with experimental results of other algorithms,it shows that the new algorithm can effectively improve the tracking accuracy of a fast moving target,and have strong robustness.
出处
《计算机工程与设计》
CSCD
北大核心
2012年第6期2277-2280,2285,共5页
Computer Engineering and Design
基金
国家自然科学基金项目(60773053)
山东省优秀中青年科学家科研奖励基金项目(2008BS01026)
关键词
目标跟踪
均值漂移
轨迹预测
最小二乘法
鲁棒性
target tracking
Mean shift
path prediction
least squares method
robustness
