摘要
红外制导技术是武器制导领域研究的热点和主要方向。针对探测跟踪过程中核相关跟踪算法(Kernelized Correlation Filter,KCF)对快速运动和严重遮挡目标的跟踪精度下降问题,提出一种融合卡尔曼滤波和运动模型的改进核相关目标跟踪算法。该算法首先利用运动模型对目标的位置进行初始估计,提出一种自适应搜索区域选择的方法。针对测试样本的置信度响应图呈现多峰平坦的情况,本文提出了一种用于目标相似度量的组合置信度测量策略,采用相关峰的锐度和置信图的平滑度约束来进一步计算疑似区域的置信度,提升算法的抗干扰能力;同时,本文也提出了一种基于最优置信度的自适应参数更新,增强模型的泛化能力。大量的仿真实验结果表明本文所提的算法的跟踪性能超过传统的核相关跟踪算法,对复杂的跟踪场景具有更强的鲁棒性与抗干扰能力。
Infrared guidance technology is gaining popularity in the field of weapon guidance and has gradually become an important method for precision guidance. A novel improved kernelized correlation tracking algorithm based on Kalman filtering and motion models is proposed to improve tracking accuracy that can deteriorated by fast motion and severe occlusion. First, an adaptive search region location method is proposed, where the optimal position is estimated by the uncertainty theory of the motion model to define the optimal search window. Since the confidence map for the test samples show a multi-peak-flat, a confidence strategy for similarity measurement is also proposed. The confidence of the correlation peak can be calculated using the sharpness of the correlation peak and a smoothness constraint. Finally, the optimal confidence is introduced to obtained an adaptive update model. A large number of simulations show that our proposed algorithm exhibits more robust performance and anti-interference ability than the traditional KCF algorithm.
作者
陈婧
孙玉娟
周万军
CHEN Jing;SUN Yujuan;ZHOU Wanjun(College of Information and Electrical Engineering,Ludong University,Yantai 264000,China;Aeronautical Foundation College,Naval Aeronautics and Astronautics University,Yantai 264001,China)
出处
《红外技术》
CSCD
北大核心
2018年第11期1106-1111,共6页
Infrared Technology
基金
鲁东大学博士科研启动基金项目(LY2013002)
关键词
目标跟踪
卡尔曼滤波
运动模型
核相关跟踪
置信度测量
不确定性理论
object tracking
Kalman filter
motion model
Kernelized correlation tracking
confidence measurement
uncertainty theory
