计算机工程 ›› 2011, Vol. 37 ›› Issue (10): 19-22.doi: 10.3969/j.issn.1000-3428.2011.10.006

• 博士论文 • 上一篇    下一篇

基于边缘化粒子滤波的红外小目标检测跟踪

温 奇 1,秦思娴 2,3,马建文 2   

  1. (1. 民政部国家减灾中心,北京 100866;2. 中国科学院对地观测与数字地球科学中心,北京 100101;3. 中国科学院研究生院,北京 100049)
  • 出版日期:2011-05-20 发布日期:2011-05-20
  • 作者简介:温 奇(1983-),男,博士研究生,主研方向:红外弱小目标检测跟踪算法;秦思娴,硕士研究生;马建文,研究员、博士生导师
  • 基金项目:
    国家自然科学青年基金资助项目(40901234);中国科学院知识创新工程青年人才领域前沿项目专项基金资助项目

Detection Track for Infrared Small Target Based on Marginalized Particle Filter

WEN Qi 1, QIN Si-xian 2,3, MA Jian-wen 2   

  1. (1. National Disaster Reduction Center of Ministry of Civil Affairs, Beijing 100866, China; 2. Center for Earth Observation and Digital Earth, Chinese Academy of Sciences, Beijing 100101, China; 3. Graduate University of Chinese Academy of Sciences, Beijing 100049, China)
  • Online:2011-05-20 Published:2011-05-20

摘要:

针对红外探测系统中单帧红外图像中低信噪比小目标检测问题,提出一种基于边缘化粒子滤波算法的检测前跟踪方法。该方法根据混合状态滤波的思想,直接利用原始图像数据,采用粒子数确定的持续概率密度函数和新生概率密度函数,推导出目标存在的概率。对没有出现在量测方程中的线性状态变量边缘化,用卡尔曼滤波器进行时间更新。实验结果证明,该方法能够减少目标跟踪的均方根误差,提高目标检测率,对低信噪比目标非常有效。

关键词: 边缘化粒子滤波, 检测前跟踪, 红外小目标, 低信噪比

Abstract: A Track Before Detection(TBD) method based on marginalized particle filter is proposed for the single-frame low Signal to Noise Ratio (SNR) infrared small target detection and tracking in infrared detection system. Under the idea of mixture state filter, the method is directly used on the original image data. The number of continue probability density particles and new birth probability density particles are deterministic and the probability of the existence of the target is derived. Linear variables disappeared in the measurement equation are marginalized and updated by Kalman filter. Experimental results prove that this method can effectively reduce the root mean square error of target tracking and improve target detection rate, especially for low SNR target.

Key words: marginalized particle filter, Track Before Detection(TBD), infrared small target, low Signal to Noise Ratio(SNR)

中图分类号: