引用本文:陈辉,杜金瑞,韩崇昭.自适应不规则形状扩展目标跟踪算法[J].控制理论与应用,2018,35(8):1111~1119.[点击复制]
CHEN Hui,DU Jin-rui,HAN Chong-zhao.An adaptive tracking algorithm for irregular shape extended target[J].Control Theory and Technology,2018,35(8):1111~1119.[点击复制]
自适应不规则形状扩展目标跟踪算法
An adaptive tracking algorithm for irregular shape extended target
摘要点击 3088  全文点击 1276  投稿时间:2017-07-17  修订日期:2018-02-07
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  10.7641/CTA.2018.70484
  2018,35(8):1111-1119
中文关键词  扩展目标跟踪  不规则形状  随机超曲面模型  自适应轮廓  拟Jaccard距离
英文关键词  extended target tracking  irregular shape  random hypersurface model  adaptive contour  quasi-Jaccard distance
基金项目  国家自然科学基金项目(61370037, 61650304, 61763029)资助.
作者单位E-mail
陈辉* 西安交通大学兰州理工大学 huich78@hotmail.com 
杜金瑞 兰州理工大学  
韩崇昭 西安交通大学  
中文摘要
      针对具有不规则形状的扩展目标跟踪问题, 在目标先验形状未知和运动过程中目标形状发生突变的情况下, 提出基于星凸形随机超曲面模型的自适应轮廓扩展目标跟踪算法. 首先, 研究星凸形扩展目标形状的径向函数傅里叶系数自适应设定问题, 基于中心轮廓距离提出不规则形状的自适应方法. 其次, 针对扩展目标在运动过程中可能发生的形状突变问题, 结合滑动窗口构造检测统计量对目标形变进行实时检测, 并提出自适应轮廓快速逼近算法对形状突变的扩展目标进行实时跟踪. 然后, 提出一种拟Jaccard距离作为形状估计质量的评价指标. 最后, 仿真实验验证了本文所提算法的有效性.
英文摘要
      In view of the extended target tracking with irregular shape, this paper proposes an adaptive contour algorithm based on star-convex random hypersurface model in the case of unknown target priori shape and shape change in the process of target evolution. First, studying the adaptive setting of Fourier coefficient of radial function, which is utilized to describe the star-convex shape, an adaptive method for irregular shape is proposed based on the centroid contour distance. Then, aiming at the sudden shape change in the process of target evolution, this paper uses the sliding window to formulate detection statistics and proposes a real-time detection method. Furthermore, an adaptive contour algorithm which can quickly approximate the real target shape is proposed to track extended target with shape change in real time. Moreover, this paper proposes the quasi-Jaccard distance to evaluate the performance of shape estimation. Simulation results verify the effectiveness of the proposed algorithm.