计算机工程与应用 ›› 2012, Vol. 48 ›› Issue (32): 191-195.

• 图形、图像、模式识别 • 上一篇    下一篇

基于形状上下文与粒子滤波的机翼跟踪

侯建波1,2,顾宏斌1,2,孙  瑾1,2   

  1. 1.南京航空航天大学 民航学院,南京 210016
    2.南京航空航天大学 飞行模拟与先进培训工程技术研究中心,南京 210016
  • 出版日期:2012-11-11 发布日期:2012-11-20

Wingtip tracking algorithm based on shape context and particles filter

HOU Jianbo1,2, GU Hongbin1,2, SUN Jin1,2   

  1. 1.School of Civil Aviation, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China
    2.Engineering Research Center for Flight Simulation and Advanced Training, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China
  • Online:2012-11-11 Published:2012-11-20

摘要: 针对飞机地面牵引移动过程中存在的潜在危险,研究了一种基于视频的翼尖跟踪方法。针对机场光照、背景变化等复杂外界情况,为克服常规图像目标跟踪方法的不足,通过简化和改进形状上下文描述翼尖轮廓特征,以粒子滤波为基本框架实现翼尖跟踪,并根据机翼的不同运动速度自适应调节状态转移方程参数,增加有效粒子个数,避免粒子衰竭。实验结果表明,该方法能够有效跟踪翼尖,具有较好的跟踪效果和鲁棒性。

关键词: 目标跟踪, 形状上下文, 粒子滤波

Abstract: To the potential dangerous existing in airplane ground movement of the airport, a wingtip tracking method which is based on video image processing and combined with conventional target tracking is proposed in this paper. Due to the complexity of illumination, background and the gesture of wingtip, this paper presents a new tracking algorithm under particles filter framework which adopts improved shape context as the tracking feature. According to the different movement speed of wing, it regulates the parameters of state transition equation adaptively, and increases the effective particles number to avoid the degeneracy problem in particle filtering. The testing results show that the method proposed in this paper has an accurate and robust performance in wingtip tracking.

Key words: object tracking, shape context, particles filter