Tracking of infrared dim small target in complex sky background
-
摘要: 为解决传统跟踪算法不能有效区分复杂天空云层背景边缘和红外弱小目标,从而在跟踪过程中产生“偏移”的问题。在时空上下文原理基础上分析跟踪“偏移”的原因,引入高斯曲率滤波,提出一种改进的时空上下文红外弱小目标跟踪算法。该算法首先采用高斯曲率滤波对上下文区域进行预处理,在保留上下文区域背景边缘的同时剔除高频的红外弱小目标和噪声,从而获得准确的红外弱小目标置信图,利用红外弱小目标置信图估计出红外弱小目标位置。采用四组复杂天空背景下的红外弱小目标图像序列进行实验,并与经典的模板匹配算法、基于粒子滤波的均值漂移算法和快速压缩跟踪算法三种跟踪算法作比较。实验结果表明,算法在主观视觉和客观评价指标方面均优于其他三种算法,具有更高的目标跟踪精度与较好的实时性,可以实现对复杂天空背景下红外弱小目标的有效跟踪。Abstract: The complex sky cloud background edge and infrared dim small target cannot be distinguished using traditional tracking algorithms, and the "deviation" problem is produced when tracking the target. To solve this problem, this paper proposes a target tracking algorithm based on Gaussian curvature filter (GCF) and spatio-temporal context (STC) after introducing the STC theory and analyzing the reason of "deviation". First of all, the context areas are dealt with GCF which can reserve the image edge and eliminate the noise and the infrared dim small target, so as to obtain the infrared dim small target's accurate confidence map. Then the infrared dim small target's coordinates are confirmed by the confidence map. Four groups of the infrared dim small target sequences in the complex sky background experimental results prove that the proposed algorithm is effective, compared with several classical algorithms, such as TMT, MS-PF and FCT. The proposed algorithm has better performance than the three traditional algorithms on subjective vision and objective evaluating indicator and shows higher tracking accuracy and better real time performance. The infrared dim small target in complex sky cloud background can be effectively tracked with this algorithm.
-
表 1 实验图像序列指标统计
Table 1. Statistics of image sequences used as candidates of experiments
image size/pixel sequence length/frame object size/pixel noise scale speed/(pixel·frame-1) motion trail moving cloud 320×320 300 3×3 large 0.5 curve thick cloud 400×400 300 3×3 very large 1 straight line plane 1 256×256 63 2×2~6×6 very large 0.4~0.8 — plane 2 570×320 375 3×3~7×7 little 0.4~0.8 — 表 2 跟踪算法性能比较
Table 2. Performance comparision of tracking algorithms
success rate(SR) average SR average FPS moving cloud thick cloud plane 1 plane 2 TMT 0.11 0.05 0 0 0.04 21 MS-PF 0.53 0.01 1 0.59 0.53 25 FCT 0.84 0.27 0.01 0.48 0.40 41 GCF-STC 1 1 0.99 0.99 0.99 54 -
[1] Liu Ruiming, Liu Yanhong, Gong Chenglong, et al. Infrared point target detection with improved template matching[J]. Infrared Physics and Technology, 2012, 55(4): 380-387. [2] 刘婧, 孙继银, 朱俊林, 等. 基于模板匹配的前视红外目标识别方法[J]. 弹箭与制导学报, 2010, 30(1): 17-19. doi: 10.3969/j.issn.1673-9728.2010.01.006Liu Jing, Sun Jiyin, Zhu Junlin, et al. FLIR target recognition algorithm based on template matching. Journal of Projectiles, Rockets Missiles and Guidance, 2010, 30(1): 17-19 doi: 10.3969/j.issn.1673-9728.2010.01.006 [3] 王晓卫, 王旭东, 贺明. 基于直方图比的背景加权的Mean Shift目标跟踪算法[J]. 强激光与粒子束, 2016, 28: 051001. doi: 10.11884/HPLPB201628.051001Wang Xiaowei, Wang Xudong, He Ming. Target tracking algorithm based on Mean Shift and histogram ratio background weighted. High Power Laser and Particle Beams, 2016, 28: 051001 doi: 10.11884/HPLPB201628.051001 [4] 魏坤, 赵永强, 潘泉, 等. 基于均值漂移和粒子滤波的红外目标跟踪[J]. 光电子·激光, 2008, 19(2): 213-217. doi: 10.3321/j.issn:1005-0086.2008.02.021Wei Kun, Zhao Yongqiang, Pan Quan, et al. IR target tracking based on mean shift and particle filter. Journal of Optoelectronics Laser, 2008, 19(2): 213-217 doi: 10.3321/j.issn:1005-0086.2008.02.021 [5] 张一帆, 田雁, 杨帆, 等. 基于改进Mean-Shift算法的红外小目标跟踪[J]. 红外与激光工程, 2014, 43(7): 2164-2169. doi: 10.3969/j.issn.1007-2276.2014.07.023Zhang Yifan, Tian Yan, Yang Fan, et al. Infrared dim target tracking based on improved mean-shift. Infrared and Laser Engineering, 2014, 43(7): 2164-2169 doi: 10.3969/j.issn.1007-2276.2014.07.023 [6] 李正周, 付红霞, 李家宁, 等. 粒子区别性稀疏表征的小弱运动目标跟踪算法[J]. 强激光与粒子束, 2016, 28: 021001. doi: 10.11884/HPLPB201628.021001Li Zhengzhou, Fu Hongxia, Li Jianing, et al. Dim moving target tracking algorithm based on particle sparse representation. High Power Laser and Particle Beams, 2016, 28: 021001 doi: 10.11884/HPLPB201628.021001 [7] Zhang Kaihua, Zhang Lei, Yang M H. Fast compressive tracking[J]. IEEE Trans Pattern Analysis and Machine Intelligence, 2014, 36(10): 2002-2015. [8] 杨德东, 蔡玉柱, 毛宁, 等. 采用核相关滤波器的长期目标跟踪[J]. 光学 精密工程, 2016, 24(8): 2037-2047. https://www.cnki.com.cn/Article/CJFDTOTAL-GXJM201608027.htmYang Dedong, Cai Yuzhu, Mao Ning, et al. Long-term object tracking based on kernelized correlation filters. Optics and Precision Engineering, 2016, 24(8): 2037-2047 https://www.cnki.com.cn/Article/CJFDTOTAL-GXJM201608027.htm [9] 邢运龙, 李艾华, 崔智高, 等. 改进核相关滤波的运动目标跟踪算法[J]. 红外与激光工程, 2016, 45: s126004. https://www.cnki.com.cn/Article/CJFDTOTAL-HWYJ2016S1037.htmXing Yunlong, Li Aihua, Cui Zhigao, et al. Moving target tracking algorithm based on improved kernelized correlation filter. Infrared and Laser Engineering, 2016, 45: s126004 https://www.cnki.com.cn/Article/CJFDTOTAL-HWYJ2016S1037.htm [10] 陈东成, 朱明, 高文, 等. 在线加权多示例学习实时目标跟踪[J]. 光学 精密工程, 2014, 22(6): 1661-1667. https://www.cnki.com.cn/Article/CJFDTOTAL-GXJM201406035.htmChen Dongcheng, Zhu Ming, Gao Wen, et al. Real-time object tracking via online weighted multiple instance learning. Optics and Precision Engineering, 2014, 22(6): 1661-1667 https://www.cnki.com.cn/Article/CJFDTOTAL-GXJM201406035.htm [11] Zhang Kaihua, Lei Zhang, Liu Qingshan, et al. Fast visual tracking via dense spatio-temporal context learning[C]//Proceedings of the European Conference on Computer Vision. 2014: 127-141. [12] 刘威, 赵文杰, 李成. 时空上下文学习长时目标跟踪[J]. 光学学报, 2016, 36: 0115001. https://www.cnki.com.cn/Article/CJFDTOTAL-GXXB201601023.htmLiu Wei, Zhao Wenjie, Li Cheng. Long-term visual tracking based on spatio-temporal context. Acta Optica Sinica, 2016, 36: 0115001 https://www.cnki.com.cn/Article/CJFDTOTAL-GXXB201601023.htm [13] 张红颖, 郑轩. 基于双目标模型的时空上下文跟踪算法[J]. 光学 精密工程, 2016, 24(5): 1215-1223. https://www.cnki.com.cn/Article/CJFDTOTAL-GXJM201605033.htmZhang Hongying, Zheng Xuan. Spatio-temporal context tracking algorithm based on dual-object model. Optics and Precision Engineering, 2016, 24(5): 1215-1223 https://www.cnki.com.cn/Article/CJFDTOTAL-GXJM201605033.htm [14] 徐建强, 陆耀. 一种基于加权时空上下文的鲁棒视觉跟踪算法[J]. 自动化学报, 2015, 41(11): 1901-1912. https://www.cnki.com.cn/Article/CJFDTOTAL-MOTO201511007.htmXu Jianqiang, Lu Yao. Robust visual tracking via weighted spatio-temporal context learning. Acta Automatica Sinica, 2015, 41(11): 1901-1912 https://www.cnki.com.cn/Article/CJFDTOTAL-MOTO201511007.htm [15] Gong Yuanhao, Sbalzarini Ivo F. Curvature filters efficiently reduce certain variational energies[J]. IEEE Trans Image Processing, 2017, 26(4): 1786-1798.