融合灰度与显著性特征的空中红外目标跟踪
|
摘要
针对红外目标特征简单且信息量少导致跟踪精度不高,提出一种基于灰度和显著性特征融合的核相关滤波算法用于空中红外目标跟踪。首先,在保证目标足够特征信息量的前提下对较大的目标进行不同等级压缩。然后将提取的二维灰度特征与显著性特征按页方式进行拼接扩展成三维特征,再将融合的特征用于核相关滤波。实验表明所提算法能够适应多种环境下的空中红外目标跟踪,跟踪精度和成功率典型值分别达到84.8%和63.9%,较大部分算法有很大提高,平均跟踪速度高达125 f/s,体现出了良好的实时性。因此,本文提出的算法在保证实时性的同时提高了跟踪的可靠性,具有一定的实用意义。
As simple feature and less information of infrared target lead to low tracking accuracy,a kernel correlation filtering algorithm based on gray and significant features fusion is proposed for aerial infrared target tracking. First of all,under the premise of ensuring enough target feature information,the larger target is compressed at different levels. Then,the extracted two-dimensional gray feature and saliency feature are spliced into three-dimensional features by page,and then the fusion features are used for kernel correlation filtering. The experimental results show that the proposed algorithm can adapt aerial infrared target tracking under a variety of environments,and the typical values of tracking accuracy and success rate reach 84. 8% and 63. 9% respectively. The average tracking speed reaches up to 125 frames per second,which reflects the good real-time. Therefore,the proposed algorithm can improve the tracking reliability and the real-time performance,and has a certain practical value.
As simple feature and less information of infrared target lead to low tracking accuracy,a kernel correlation filtering algorithm based on gray and significant features fusion is proposed for aerial infrared target tracking. First of all,under the premise of ensuring enough target feature information,the larger target is compressed at different levels. Then,the extracted two-dimensional gray feature and saliency feature are spliced into three-dimensional features by page,and then the fusion features are used for kernel correlation filtering. The experimental results show that the proposed algorithm can adapt aerial infrared target tracking under a variety of environments,and the typical values of tracking accuracy and success rate reach 84. 8% and 63. 9% respectively. The average tracking speed reaches up to 125 frames per second,which reflects the good real-time. Therefore,the proposed algorithm can improve the tracking reliability and the real-time performance,and has a certain practical value.
引文
[1]Lei L,Huang Z.Improved motion information-based infrared dim target tracking algorithms[J].Infrared Physics&Technology,2014,67:341-349.
[2]YONG Yang,WANG Shengzhe,WANG Bingxue,et al.Infrared moving target tracking based on sparse feature selection[J].Laser&Infrared,2015,45(4):446-451.(in Chinese)雍杨,王升哲,王兵学,等.采用稀疏特征选择的红外运动目标跟踪方法[J].激光与红外,2015,45(4):446-451.
[3]WANG Jun,HE Ziqing,LIU Hongyan,et al.Rapid infrared target extraction based on simplest visual saliency[J].Laser&Infrared,2016,46(7):880-884.(in Chinese)王军,何子清,柳红岩,等.基于最简视觉显著性的红外目标快速提取方法[J].激光与红外,2016,46(7):880-884.
[4]JIN Xiaohui,YANG Wei,BAI Xiaofang.A Method using infrared thermal imager to detect and track hidden[J].Infrared Technology,2014,(8):639-643.(in Chinese)金晓会,杨卫,白晓方.一种用红外热像仪检测与跟踪遮挡目标的方法[J].红外技术,2014,(8):639-643.
[5]Wei Zhong,Huchuan Lu,Ming-Hsuan Yang.Robust object tracking via sparsity-based collaborative model[C]//Computer Vision and Pattern Recognition.IEEE,2012:1838-1845.
[6]Danelljan M,Hager G,Khan F S,et al.Learning Spatially Regularized Correlation Filters for Visual Tracking[C]//IEEE International Conference on Computer Vision.IEEE Computer Society,2015:4310-4318.
[7]Bibi A,Mueller M,Ghanem B.Target Response Adaptation for Correlation Filter Tracking[C]//European Conference on Computer Vision.Springer International Publishing,2016:419-433.
[8]Bibi,Adel Aamer,Ghanem,Bernard.Multi-template Scale-Adaptive Kernelized Correlation Filters[C]//IEEE International Conference on Computer Vision Workshop.IEEE,2015:613-620.
[9]Li Y,Zhu J.A Scale Adaptive Kernel Correlation Filter Tracker with Feature Integration[C]//European Conference on Computer Vision.Springer,Cham,2014:254-265.
[10]Henriques J F,Rui C,Martins P,et al.High-Speed Tracking with Kernelized Correlation Filters[J].IEEE Transactions on Pattern Analysis&Machine Intelligence,2014,37(3):583-596.
[11]PAN Zhenfu,ZHU Yongli.Kernelized Correlation Filters Object Tracking Method with Multi-Scale Estimation[J].Laser&Optoelectronics Progress,2016,53(10):199-205.(in Chinese)潘振福,朱永利.多尺度估计的核相关滤波器目标跟踪方法[J].激光与光电子学进展,2016,53(10):199-205.
[12]Hou X,Zhang L.Saliency Detection:A Spectral Residual Approach[C]//Computer Vision and Pattern Recognition,2007.CVPR'07.IEEE Conference on.IEEE,2007:1-8.
[13]Wu Y,Lim J,Yang M H.Online Object Tracking:A Benchmark[C]//IEEE Conference on Computer Vision and Pattern Recognition.IEEE Computer Society,2013:2411-2418.
[2]YONG Yang,WANG Shengzhe,WANG Bingxue,et al.Infrared moving target tracking based on sparse feature selection[J].Laser&Infrared,2015,45(4):446-451.(in Chinese)雍杨,王升哲,王兵学,等.采用稀疏特征选择的红外运动目标跟踪方法[J].激光与红外,2015,45(4):446-451.
[3]WANG Jun,HE Ziqing,LIU Hongyan,et al.Rapid infrared target extraction based on simplest visual saliency[J].Laser&Infrared,2016,46(7):880-884.(in Chinese)王军,何子清,柳红岩,等.基于最简视觉显著性的红外目标快速提取方法[J].激光与红外,2016,46(7):880-884.
[4]JIN Xiaohui,YANG Wei,BAI Xiaofang.A Method using infrared thermal imager to detect and track hidden[J].Infrared Technology,2014,(8):639-643.(in Chinese)金晓会,杨卫,白晓方.一种用红外热像仪检测与跟踪遮挡目标的方法[J].红外技术,2014,(8):639-643.
[5]Wei Zhong,Huchuan Lu,Ming-Hsuan Yang.Robust object tracking via sparsity-based collaborative model[C]//Computer Vision and Pattern Recognition.IEEE,2012:1838-1845.
[6]Danelljan M,Hager G,Khan F S,et al.Learning Spatially Regularized Correlation Filters for Visual Tracking[C]//IEEE International Conference on Computer Vision.IEEE Computer Society,2015:4310-4318.
[7]Bibi A,Mueller M,Ghanem B.Target Response Adaptation for Correlation Filter Tracking[C]//European Conference on Computer Vision.Springer International Publishing,2016:419-433.
[8]Bibi,Adel Aamer,Ghanem,Bernard.Multi-template Scale-Adaptive Kernelized Correlation Filters[C]//IEEE International Conference on Computer Vision Workshop.IEEE,2015:613-620.
[9]Li Y,Zhu J.A Scale Adaptive Kernel Correlation Filter Tracker with Feature Integration[C]//European Conference on Computer Vision.Springer,Cham,2014:254-265.
[10]Henriques J F,Rui C,Martins P,et al.High-Speed Tracking with Kernelized Correlation Filters[J].IEEE Transactions on Pattern Analysis&Machine Intelligence,2014,37(3):583-596.
[11]PAN Zhenfu,ZHU Yongli.Kernelized Correlation Filters Object Tracking Method with Multi-Scale Estimation[J].Laser&Optoelectronics Progress,2016,53(10):199-205.(in Chinese)潘振福,朱永利.多尺度估计的核相关滤波器目标跟踪方法[J].激光与光电子学进展,2016,53(10):199-205.
[12]Hou X,Zhang L.Saliency Detection:A Spectral Residual Approach[C]//Computer Vision and Pattern Recognition,2007.CVPR'07.IEEE Conference on.IEEE,2007:1-8.
[13]Wu Y,Lim J,Yang M H.Online Object Tracking:A Benchmark[C]//IEEE Conference on Computer Vision and Pattern Recognition.IEEE Computer Society,2013:2411-2418.