摘要
针对棒材在线计数系统中多运动目标的跟踪问题,提出了一种利用棒材端面二值图像的垂直投影曲线的运动特征、几何特征和位置特征来实现快速、稳定跟踪的算法。它能够在线自适应选择跟踪窗口的位置和大小,然后依据距离相关度、面积相似度和置信度函数评价每次的观测数据,使得跟踪过程具有良好的容错性。现场应用结果表明:该跟踪算法实时性好、可靠性高,保证了棒材在线计数的准确性。
To solve the problem of multi-moving targets tracking in the online rebar counting system, a fast and stable online rebars tracking algorithm is proposed,which fuses the characteristics of move- ment, geometry and location in the vertical projection curve derived from the rebar end binary im- age. The algorithm can select the tracking window position and size online adaptively, and evaluate observational data in accordance with the distance correlative degree, area similar degree and the confidence degree, which allows the tracking process a good fault tolerance. The application results on the spot show that the tracking algorithm is in good real-time, highly reliable and can ensure the accuracy of online rebars counting.
出处
《福建工程学院学报》
CAS
2013年第6期517-521,526,共6页
Journal of Fujian University of Technology
基金
福建省自然科学基金计划项目(2012D069)
福建省教育厅资助项目(JA13217)
关键词
目标跟踪
多特征融合
距离相关度
面积相似度
置信度
target tracking
multi-feature fusion
distance correlative degree
area similar degree
confidence degree
