多帧数据关联的杂波红外弱小目标跟踪方法研究
|
摘要
为解决杂波环境中红外弱小目标遮挡、丢失和跟踪虚警问题,提高制导控制系统的稳定性,对一种自反馈的点目标跟踪方法进行了研究。用启发式数据关联方法,将点目标跟踪的两帧关联问题扩展到多帧时间维度。用运动一致和辐射一致两个约束条件确立当前帧目标及前一帧目标间的关联关系,构造有向图,将多帧多目标的映射问题转换为加权有向图的最大边覆盖问题。计算有向边映射代价,寻找加权有向图最大匹配,修正前一帧目标映射关系,建立前一帧与当前帧目标的映射关系。关于成像导引头灵敏阈和海面闪烁条件下目标跟踪的仿真结果表明:所提方法通过对多帧数据关联的局部最优化求解,具处理目标短时遮挡、丢失和虚警的能力,能在杂波环境中快速识别出红外弱小目标,跟踪稳定有效。
To solve the weak and small infrared target under clutter environment occlusion,loss and trace false alarm problem and improve the stability of the guidance and control system,a target tracking method for feedback was proposed in this paper.Using a heuristic method of data association,tracking two frames associated problems was extended to multiple frames on a time dimension.The association between the present frame and the latest former frame was established by two constrained conditions which were motion consistent and radiation consistent.The directed graph was constructed.The mapping of multiply frames multiply targets was transferred to maximum edge coverings of weighted directed graph.The cost of the directed edge was calculated to search the maximum matching of the graph,which was used for correcting the mapping of the latest former frame and establishing the mapping of the present frame and the latest former frame.The target tracking simulations of the seeker sensitivity threshold and sea surface scintillation were carried out.The results show that the proposed method has the ability to deal with the short-block,missing and false alarms through the local optimal solution of multi-frame data association and can identify the infrared weak and small target in clutter environment.The tracking is stable and effective.
To solve the weak and small infrared target under clutter environment occlusion,loss and trace false alarm problem and improve the stability of the guidance and control system,a target tracking method for feedback was proposed in this paper.Using a heuristic method of data association,tracking two frames associated problems was extended to multiple frames on a time dimension.The association between the present frame and the latest former frame was established by two constrained conditions which were motion consistent and radiation consistent.The directed graph was constructed.The mapping of multiply frames multiply targets was transferred to maximum edge coverings of weighted directed graph.The cost of the directed edge was calculated to search the maximum matching of the graph,which was used for correcting the mapping of the latest former frame and establishing the mapping of the present frame and the latest former frame.The target tracking simulations of the seeker sensitivity threshold and sea surface scintillation were carried out.The results show that the proposed method has the ability to deal with the short-block,missing and false alarms through the local optimal solution of multi-frame data association and can identify the infrared weak and small target in clutter environment.The tracking is stable and effective.
引文
[1]高璐,张大志,田金文.红外序列图像目标跟踪的自适应Kalman滤波方法[J].红外与激光工程,2007,36(5):729-732+757.
[2]万琴,王耀南.基于卡尔曼滤波器的运动目标检测与跟踪[J].湖南大学学报(自然科学版),2007,34(3):36-40.
[3]KHAN Z,BALCH T,DELLAERT F.MCMCbased particle filtering for tracking a variable number of interacting targets[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2005,27(11):1805-1819.
[4]马兆男,裴腾达,张浩.一种改进的粒子滤波目标跟踪算法[J].辽宁工程技术大学学报(自然科学版),2016,35(9):978-982.
[5]SALARI V,SETHI I K.Feature point correspondence in the presence of occlusion[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1990,12(1):97-91.
[6]RANGARAJAN K,SHA M.Establishing motion correspondence[J].Image Understanding,1991,24(6):56-73.
[7]CHETVERIKOV D,VERESTOY J.Feature point tracking for incomplete trajectories[J].Nerucomputing,1999,6(2):21-338.
[8]VEENMAN C J,REINDERS M J T,BACKER E.Resolving motion correspondence for densely moving points[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2001,23(1):54-72.
[9]SHAFIQUE K,SHAH M.A noniterative greedy algorithm for multiframe point correspondence[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2005,27(1):51-65.
[10]KUEN J,LIM K M,LEE C P.Self-taught learning of a deep invariant representation for visual tracking via temporal slowness principle[J].Pattern Recognition,2015,48(10):2964-2982.
[11]TOROSLU I H,UCOLUK G.Incremental assignment problem[J].Information Science,2007:1523-1529.
[12]LAVAGETTO F.Automatic target detection in infrared sequences through semantic labeling[J].SPIE,1990,1349:476-485.
[13]刘颖彬.基于DSP和FPGA的红外小目标检测系统[J].电子技术与软件工程,2016,23(3):36-40.
[2]万琴,王耀南.基于卡尔曼滤波器的运动目标检测与跟踪[J].湖南大学学报(自然科学版),2007,34(3):36-40.
[3]KHAN Z,BALCH T,DELLAERT F.MCMCbased particle filtering for tracking a variable number of interacting targets[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2005,27(11):1805-1819.
[4]马兆男,裴腾达,张浩.一种改进的粒子滤波目标跟踪算法[J].辽宁工程技术大学学报(自然科学版),2016,35(9):978-982.
[5]SALARI V,SETHI I K.Feature point correspondence in the presence of occlusion[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1990,12(1):97-91.
[6]RANGARAJAN K,SHA M.Establishing motion correspondence[J].Image Understanding,1991,24(6):56-73.
[7]CHETVERIKOV D,VERESTOY J.Feature point tracking for incomplete trajectories[J].Nerucomputing,1999,6(2):21-338.
[8]VEENMAN C J,REINDERS M J T,BACKER E.Resolving motion correspondence for densely moving points[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2001,23(1):54-72.
[9]SHAFIQUE K,SHAH M.A noniterative greedy algorithm for multiframe point correspondence[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2005,27(1):51-65.
[10]KUEN J,LIM K M,LEE C P.Self-taught learning of a deep invariant representation for visual tracking via temporal slowness principle[J].Pattern Recognition,2015,48(10):2964-2982.
[11]TOROSLU I H,UCOLUK G.Incremental assignment problem[J].Information Science,2007:1523-1529.
[12]LAVAGETTO F.Automatic target detection in infrared sequences through semantic labeling[J].SPIE,1990,1349:476-485.
[13]刘颖彬.基于DSP和FPGA的红外小目标检测系统[J].电子技术与软件工程,2016,23(3):36-40.