多伯努利滤波的快速红外弱小目标检测与跟踪

doi:10.3969/j.issn.1001-2400.2016.04.013

摘要/Abstract

摘要：

针对复杂背景下多个弱目标检测与跟踪中存在的跟踪不稳定、非实时及量测模型高度非线性问题，提出一种基于多伯努利滤波的快速检测与跟踪算法．首先，采用改进的Robinson Guard算法抑制背景杂波，避免强起伏背景图像中目标被抑制的问题；其次，采用平方根容积卡尔曼滤波实现多伯努利检测前跟踪，在保证实时跟踪的同时，解决了滤波的高度非线性，避免了协方差矩阵负定造成的数值不稳定．实测红外背景图像实验表明，改进的Robinson Guard算法能够有效抑制背景杂波、保留弱目标信息，平方根容积卡尔曼多伯努利检测前跟踪能更准、更稳定地估计目标数目和状态，实现目标的实时检测与跟踪．

关键词: 红外图像序列, 背景杂波抑制, Robinson Guard滤波, 平方根容积卡尔曼, 多伯努利滤波, 检测前跟踪

Abstract:

In view of the problems of tracking instability and non real-time and highly nonlinear measurement model for the detection and tracking of multiple Infrared(IR) dim targets under the complicated background, a fast detection and tracking algorithm based on multi-Bernoulli filter is presented. Firstly, the improved Robinson Guard filter suppresses the background clutter in order to avoid the suppression of the target in the strong background image. Then, the square-root cubature Kalman filter (SCK) performs the multi-Bernoulli track-before-detect (MB-TBD). It can perform real-time tracking in the highly nonlinear model. The numerical instability caused by a negative definite covariance matrix is avoided. The experiments with some real IR background images show that the improved Robinson Guard algorithm can suppress the background clutter effectively and preserve the dim target information. It is also shown that SCK-MB-TBD can estimate the number and states of targets accurately and stably so as to complete target detection and tracking.

李翠芸;李宁;姬红兵. 多伯努利滤波的快速红外弱小目标检测与跟踪[J]. 西安电子科技大学学报, 2016, 43(4): 69-74+190.

LI Cuiyun;LI Ning;JI Hongbing. Fast IR dim small targets detection and tracking based on the multi-Bernoulli filter[J]. Journal of Xidian University, 2016, 43(4): 69-74+190.

参考文献

［1］ LIU L, HUANG Z J. Infrared Dim Target Detection Technology Based on Background Estimate ［J］. Infrared Physics ＆ Technology, 2014, 62: 59-64.
［2］林涛, 韩平丽, 刘飞. 利用窄波段像素色比的红外弱小目标检测方法［J］. 西安电子科技大学学报, 2014, 41(4): 58-63, 157.
LIN Tao, HAN Pingli, LIU Fei. Small Target Detection Utilizing the Narrow Wavebands ［J］. Journal of Xidian University, 2014, 41(4): 58-63, 157.
［3］武斌, 李鹏. 一种新的红外弱小目标检测前跟踪算法［J］. 西安电子科技大学学报, 2011, 38(3): 107-113.
WU Bin, LI Peng. Novel Track-before-detect Algorithm for Small Infrared Target ［J］. Journal of Xidian University, 2011, 38(3): 107-113.
［4］ CAI F, FAN HQ, FU Q. Track-before-detect Filter for Sensors with Point Spread Function Using Thresholded Measurements ［C］//Proceedings of 17th International Conference on Information Fusion. Piscataway: IEEE, 2014: 1-6.
［5］罗军辉, 姬红兵, 刘靳. 一种基于空间滤波的红外小目标检测算法及其应用［J］. 红外与毫米波学报, 2007, 26(3): 209-212.
LUO Junhui, JI Hongbing, LIU Jin. Algorithm of IR Small Targets Detection Based on Spatial Filter and Its Application［J］. Journal of Infrared and Millimeter Waves, 2007, 26(3): 209-212.
［6］ FARAJZADEH M, MAHMOODI A, ARVAN M R, et al. Detection of Small Target Based on Morphological Filters ［C］//Proceedings of 20th Iranian Conference on Electrical Engineering. Washington: IEEE, 2012: 1097-1101.
［7］ MAHLER R. Multitarget Bayes Filtering via First-order Multitarget Moments［J］. IEEE Transactions on Aerospace and Electronic Systems, 2003, 39(4): 1152-1178.
［8］ MAHLER R. PHD Filters of Higher Order in Target Number［J］. IEEE Transactions on Aerospace and Electronic Systems, 2007, 43(4): 1523-1543.
［9］ MAHLER R. Statistical Multisource-multitarget Information Fusion.Boston［M］. London: Artech House Publishers, 2007.
［10］李翠芸, 江舟, 姬红兵. 一种新的未知杂波环境下的PHD滤波器［J］. 西安电子科技大学学报, 2014, 41(5): 18-23.
LI Cuiyun, JIANG Zhou, JI Hongbing. Novel PHD Filter in Unknown Clutter Environment［J］. Journal of Xidian University, 2014, 41(5): 18-23.
［11］PUNITHAKUMAR K, KIRUBARAJAN T, SINHA A. A Sequential Monte Carlo Probability Hypothesis Density Algorithm for Multitarget Track-before-detect［C］//Proceedings of the SPIE: 5913. Bellingham: SPIE, 2005: 1-8.
［12］ VO B N, VO B T, PHAM N T, et al. Joint Detection and Estimation of Multiple Objects from Image Observation［J］. IEEE Transactions on Signal Processing, 2010, 58(10): 5129-5140.
［13］林再平, 周一宇, 安玮. 基于势概率假设密度滤波的检测前跟踪算法［J］. 红外与毫米波学报, 2013, 32(5): 437-443.
LIN Zaiping, ZHOU Yiyu, AN Wei. Track-before-detect Algorithm Based on Cardinalized Probability Hypothesis Density Filter［J］. Journal of Infrared and Millimeter Waves, 2013, 32(5): 437-443.
［14］李翠芸, 曹潇男, 廖良雄. 高斯粒子PHD滤波的多个弱小目标TBD算法［J］. 系统工程与电子技术, 2015, 37(4): 740-745.
LI Cuiyun, CAO Xiaonan, LIAO Liangxiong. Track before Detect Using Gaussian Particle Probability Hypothesis Density［J］. Systems Engineering and Electronics, 2015, 37(4): 740-745.
［15］ LI C Y, JI H B, ZOU Q B, et al. A Novel Multi-Bernoulli Filter for Joint Target Detection and Tracking ［C］//Proceedings of 2013 IET International Conference on Information and Communications Technologies. Stevenage: IET, 2013: 176-180.
［16］ VO B T, VO B N, CANTONI A. The Cardinality Balanced Multitarget Multi-Bernoulli Filter and Its Implementations［J］. IEEE Transactions on Signal Processing, 2009, 57(2): 409-423.
［17］郝燕玲, 杨峻巍, 陈亮, 等. 平方根容积卡尔曼滤波器［J］. 弹箭与制导学报, 2012, 32(2): 169-172.
HAO Yanling, YANG Junwei, CHEN Liang, et al. Square Root Cubature Kalman Filter［J］. Journal of Projectiles,Rockets,Missiles and Guidance, 2012, 32(2): 169-172.
［18］ SCHUHMACHER D, VO B T, VO B N. A Consistent Metric for Performance Evaluation of Multi-object Filters［J］. IEEE Transactions on Signal Processing, 2008, 56(8): 3447-3457.

[1]	张佳琦,陶海红,张修社,韩春雷. 一种利用量测空间聚类的多帧检测前跟踪算法[J]. 西安电子科技大学学报, 2021, 48(5): 231-238.
[2]	李翠芸,李洋,姬红兵,石仁政. 一步数据关联GLMB扩展目标跟踪算法[J]. 西安电子科技大学学报, 2020, 47(5): 137-143.
[3]	陈帅霖;罗丰;张林让;胡冲;陈世超. 采用动态规划的机动目标检测前跟踪算法[J]. 西安电子科技大学学报, 2017, 44(5): 39-45.
[4]	王海环;王俊. 一种改进的多伯努利多目标跟踪算法[J]. 西安电子科技大学学报, 2016, 43(6): 176-182.
[5]	何嘉懿;廖桂生;杨志伟;孙存光. 一种参数化轨迹增强的TBD算法[J]. J4, 2015, 42(6): 17-22.
[6]	张鹏;张林让;胡子军. HPRF雷达距离延拓检测前跟踪方法[J]. J4, 2014, 41(5): 207-212.
[7]	夏双志;严俊坤;戴奉周;刘宏伟. 目标存在状态变量的非齐次马尔可夫链模型[J]. J4, 2013, 40(4): 48-55.
[8]	武斌；李鹏. 一种新的红外弱小目标检测前跟踪算法[J]. J4, 2011, 38(3): 107-113.
[9]	吴兆平;朱凯然;苏涛;卢锦. 采用改进粒子滤波的雷达扩展目标检测前跟踪[J]. J4, 2011, 38(2): 99-104.
[10]	王建涛. 一种双波段红外图像弱小目标融合检测新算法 [J]. J4, 2008, 35(3): 530-535.
[11]	刘靳;姬红兵. 基于移动式加权管道滤波的红外弱小目标检测 [J]. J4, 2007, 34(5): 743-747.