红外序列图像目标轨迹关联检测算法研究与硬件设计
|
摘要
随着精确制导技术的发展和红外技术的进步,红外成像制导技术越来越受到研究人员的重视,红外序列图像中弱小目标检测技术是红外成像系统中的核心技术之一,研究运算量小、性能高、利于硬件实时实现的检测算法是当今和今后的一个热门问题。
本文分析研究了一种适合硬件实现的序列图像目标轨迹关联检测算法。该算法在传统的轨迹关联算法基础上进行了改进,采用两步处理的方式实现目标检测。首先根据目标在序列图像中的运动特性,采用直线判别准则对目标与噪声点进行分离,然后利用最小二乘预测器对剩下的候选目标点进行轨迹预测,并对预测的轨迹进行匹配及置信度判断,最终检测出目标。
本文在详细论述算法的基础上,提出了算法的整体结构,并根据实时性要求,对整个算法进行了FPGA硬件设计。文中详细论述了候选目标点筛选模块、轨迹预测模块、轨迹匹配模块等模块的设计思想及其实现方法,重点讲述了移位相加单元实现除法运算的方法,并围绕精确度问题进行了详细讨论且给出了解决方案。
本文采用面向可综合的寄存器传输级设计方法完成电路设计,利用ModelSim仿真工具验证了设计的正确性。从综合结果看出,本文所提出的序列图像目标轨迹关联检测算法满足实时性要求,且消耗的硬件资源少,为红外弱小目标检测系统在单片FPGA上实现打下了良好的基础,对红外成像制导系统整体性能的提高有着重要的现实意义。
With the development of precision guidance technique and infrared technique, researchers are paying more and more attention to infrared imaging guidance technique.The technique of small infrared target detection is one of the key technologies in the system of infrared image information.So researching the detection algorithm of small calculation,high performance and less logic resource has been the subject of intense investigaion now and in the future.
In this paper, a track-assosiation detection algorithm in IR image sequeences was studied which could be realized by hardware conveniently. Based on the traditional track-associated algorithm, this algorithm was made from two steps: Firstly, According to continuity and relativity of movement,targets were easily separated from noise by line-rule; secondly, the detection of target was performed by predicting target position by least-square algorithm and gliding confidence test.
Based on the theoretically analyzed, in this paper the algorithm’s archtecture was presented. And in order to satisfy real-time requirement, an FPGA hardware realization solution to the detection algorithm was proposed. The main idea and realization of several modules such as target-filtration module, track-prediction module and track- macthing module, had been described in detail. Especially, the shift-add cell was analyzed which was used to realize multiplication and division in hardware. At the same time, the precision problem of the shift-add cell was discussed and also a solution was given .
The hardware realization had been done with register. The validity of the design was verified by the ModelSim simulator. Theoretical analysis and simulations indicate that the design meets the real-time requirement and has simple structrue and less logic resource.This method make a solid foundation for the realization of infrared detection system on one piece of FPGA, and also make a sense of the improvement of infrared imaging system’s performance.
本文分析研究了一种适合硬件实现的序列图像目标轨迹关联检测算法。该算法在传统的轨迹关联算法基础上进行了改进,采用两步处理的方式实现目标检测。首先根据目标在序列图像中的运动特性,采用直线判别准则对目标与噪声点进行分离,然后利用最小二乘预测器对剩下的候选目标点进行轨迹预测,并对预测的轨迹进行匹配及置信度判断,最终检测出目标。
本文在详细论述算法的基础上,提出了算法的整体结构,并根据实时性要求,对整个算法进行了FPGA硬件设计。文中详细论述了候选目标点筛选模块、轨迹预测模块、轨迹匹配模块等模块的设计思想及其实现方法,重点讲述了移位相加单元实现除法运算的方法,并围绕精确度问题进行了详细讨论且给出了解决方案。
本文采用面向可综合的寄存器传输级设计方法完成电路设计,利用ModelSim仿真工具验证了设计的正确性。从综合结果看出,本文所提出的序列图像目标轨迹关联检测算法满足实时性要求,且消耗的硬件资源少,为红外弱小目标检测系统在单片FPGA上实现打下了良好的基础,对红外成像制导系统整体性能的提高有着重要的现实意义。
With the development of precision guidance technique and infrared technique, researchers are paying more and more attention to infrared imaging guidance technique.The technique of small infrared target detection is one of the key technologies in the system of infrared image information.So researching the detection algorithm of small calculation,high performance and less logic resource has been the subject of intense investigaion now and in the future.
In this paper, a track-assosiation detection algorithm in IR image sequeences was studied which could be realized by hardware conveniently. Based on the traditional track-associated algorithm, this algorithm was made from two steps: Firstly, According to continuity and relativity of movement,targets were easily separated from noise by line-rule; secondly, the detection of target was performed by predicting target position by least-square algorithm and gliding confidence test.
Based on the theoretically analyzed, in this paper the algorithm’s archtecture was presented. And in order to satisfy real-time requirement, an FPGA hardware realization solution to the detection algorithm was proposed. The main idea and realization of several modules such as target-filtration module, track-prediction module and track- macthing module, had been described in detail. Especially, the shift-add cell was analyzed which was used to realize multiplication and division in hardware. At the same time, the precision problem of the shift-add cell was discussed and also a solution was given .
The hardware realization had been done with register. The validity of the design was verified by the ModelSim simulator. Theoretical analysis and simulations indicate that the design meets the real-time requirement and has simple structrue and less logic resource.This method make a solid foundation for the realization of infrared detection system on one piece of FPGA, and also make a sense of the improvement of infrared imaging system’s performance.
引文
[1]黄世奇,禹春来,刘代志等.成像精确制导技术分析与研究[J].导弹与航天运载技术,2005(5)
[2]杨树谦.精确制导技术发展现状与展望[J].航天控制,2004.22(4)
[3]曾宪林.红外成像导引头及其成像制导武器评述[J].航天电子对抗,2004(5)
[4]熊辉红.外成像制导信息处理算法研究[D].国防科学技术大学博士学位论文,1998
[5]俞建成,孙胜利,陈桂林.红外图像中小目标检测研究[J].红外,2006.27(4)
[6] Zhang Bing.The predicting and matching detection algorithm of moving point target in image sequences [A] . IEEE International Conference on Neural Networks & Signal Processing[C].NanJing,China ICNNSP,2003
[7]孙德宝,周卫祥,彭家雄.红外图像序列运动小目标检测的预处理算法研究[J].红外与激光工程, 2000.29(2):12~14
[8]杨卫平,沈振康.红外图像序列小目标检测预处理技术[J].红外与激光工程,1998,2.27(1):23~28
[9]徐英.红外图像序列中运动弱小目标检测的方法研究[J].红外技术,2002,11.24(6):27~30
[10] Zang Hong,Zhao Bao jun,Mao Er-ke.A real-time effective method for infrared point target detection in spatially varying clutter[A].IEEE,CIE International Conference on radar,Proceedings [C].2001,461~464
[11] Muller M.Saliency Measures in Cluttered IR Images for ATR[J].SPIE.Apr 1999,3699: 150~154
[12] Scharfand L L,Friedlander B.Matched subspaced etectors[J].IEEE Trans.On Sig.Proc.Aug 1994,42(8):2146~2156
[13] Courtney I Hilliard.Selection of a clutter rejection algorithm for real-time target detection from an air-borne platform[A].SPIE 4048,Signal and data processing of small targets[C].2000,Orlando USA
[14] Deshpande S D, Er M H,Ronda V.Max-mean and max-median filters for detection of small targets[J].SPIE Oct 1999,3809:74~83
[15] Wang Yang ,Zhen Qinbo,Zhang Junping.Real time detection of small target in IR grey image based on mathematical morphology[J].Infrared and Laser Engineering.2003,32(1):28~31
[16] Ye Bin,Peng Jiaxiong.Small target detection based on energy accumulation andorder morphology filtering in infrared image[J].Journal of Image and Graphics,2002,7(A) (3):251~255
[17] Ye Bin,Peng Jiaxiong.Moving small target detection based on order morphology filtering in infrared image sequences [J].Journal of Data Acquisition & Processing 2001,16 (3):315~319
[18]左震,张天序.基于小波变换的红外成像海面小目标检测算法[J].红外与激光工程,1998,6.27(3):42~45
[19]朱梦宇,赵保军,韩月秋.一种实用的红外弱小目标检测跟踪处理机研究[J].激光与红外,2002,12. 32(6):407~409
[20]杨威,张田文.复杂景物环境下运动目标检测的新方法[J].计算机研究与发展,1998.35(8):724~728
[21] Rauch H. E,Futterman W. I. And Kemmer D. B.Background Suppression and Tracking with a staring Mosaic sensor[J].Optical Engineering,1981,20(1):103~110
[22] Wei P,Zeidler J,Ku W.Analysis of multiframe target detecting using pixel statistics[J].IEEE Trans,Aerosp.Electron.Syst.1995,AES-31(1):238~247
[23]倪林,俞斯乐.用于运动估计的相位相关计算[A].计算机视觉与智能控制学会第二届学术会议论文集[C].武汉,1991
[24] VegaL C,Cagigal M P,Prieto P M.Determining the velocity of a moving object from a series of photon-limited frames using a temporal Fourier transfor- m[J].SPIE.Oct 1993,1954:198~203
[25] Dong Yongqiang,Tao Ran,Zhou Siyong,et al.SAR moving target detection and imaging based on WHH Transform[J] . Journal of Beijing Institute of Technology.1999,8(1):94~100
[26] Tracie C,Moore,Stephen Lowery,Performance of Various Background Estimation and Object Detection Algorithms on Real Flight Test Interceptor Seeker Data[A].Part of the SPIE Conference on Signal and Data Processing of Small Targets [J].SPIE.July 1999,3809:86~94
[27] I. S. Reed et al.Optical Moving Target Identification Algorithm for Optical Image Sequences[J].IEEE Trans.AES,1990,26(3):434-440
[28] Thompson W B,Mutch K M,Berzins V A.Detecting moving objects using the rigidity constraint[J] . IEEE Transactions on Pattern Analysis and Machine Intelligence.1993,15:162~166
[29] Nishiguchi K,Kobayashi M,Ichikawa A.Small target detection from imagesequences using recursive max filter [J].SPIE.1995,2561:153~165
[30] Diani M,Corsiniq Baldacci.A Space-time processing for the detection of air borne targets in IR image sequences[J].IEE Proceeding of Vision,Image and Signal Processing.Jun 2001,148(3):151~157
[31]彭嘉雄,彭铁.弱目标检测的图像流法[J].红外与激光工程,1996,8,25(4):34~40
[32]李国宽,李红,彭嘉雄.从图像序列中检测定位小目标的一种方法[J].华中理工大学学报.1999,8,27(8):63~65
[33] Cheung J F,Y Wicks M C,Genello G J.A Statistical Theory for Optimal Detection of Moving Objects in Variable Corruptive Noise[J].IEEE Transactions on Image Process- ing.Dec 1999,8(12):1772~1787
[34] Choi J,Jang J,Lee S,et.al.Multiple moving objecte stimation in images equences of natural scene[J].Opt Eng.1997,36(8):2176~2183
[35] Blostein S,D Richardson H S.A sequential detection approach to target tracking[J].IEEE Trans.Aerospace and Electronic Systems.1994,30(1):197~211
[36] Blostein S D,Huang T S.Detection small moving objects in image sequences using sequential hypothesis testing[J].IEEE Trans Signal Processing.1991,39(7):611~1629
[37] Collet C,Fablet R.Motion Estimation on Infrared Image Sequences[J].SPIE.1999 3699:67~77
[38] Thomas Fechner,Rainer Hach,Oliver Rockinger,Christoph Stahl.Detection and tracking of small target in aerial image sequences under unknown sensor motion[A].Part of the SPIE Conference on Signal Processing,Sensor Fusion,and Target Recognition VII[C],Orlando,Florida,April 1998,SPIE 3374:259~266
[39] Askar H,LI Xiao feng,LI Zai ming.Background clutter suppression and dim moving point targets detection using nonparametric method[A].IEEE 2002 International Conference on Communications,Circuits and Systems and West Sino Expositions [C].2002:982~986
[40] Grossman S B,Emmons R B.Performance analysis and size optimization of focal planes for point-source tracking algorithm applications[J].Optical Engineer- ing.Apr 1984,23(2):167~176
[41]熊艳,彭嘉雄,丁明跃等.一种低信噪比条件下点目标检测方法[J].电子学报.1996,12. 24(12):41~45
[42]张金锁,宋利权,靳民会等.基于带通滤波的红外小目标捕获算法研究[J].红外与激光工程,2002,8.31(4):294~296
[43] Lion R,Inigom,Azimi-Sadjad E S.Dim target detection using high order correlation method[J].IEEE Trans Aerosp,Electron,Syst.1994,AES-29(1):197~212
[44]黄林梅,基于动态规划的红外运动小目标检测算法[J].红外激光与工程,2004.33(3)
[45] Chen HuaMing,Sun Guangfu,Lu Huanzhang,Chang Qing.Multi-level thresholding and confidence testing detection algorithm for small moving target in image sequences[J].SPIE,Vol. 4473,2001:520~529
[46]周冰,王永仲,应家驹.弱小目标检测技术浅析[J].红外技术,2007.29(1)
[47]陈玉丹,周冰.红外小目标检测中的背景预处理技术研究[J].科学技术与工程,2006.6(18)
[48]李学广.红外图像预处理算法研究及硬件实现[D].哈尔滨工程大学硕士学位论文,2007
[49]方强,孙广富,卢焕章.红外焦平面图像稳像处理的FPGA实现[J].微处理机, 2001(1)
[50]蔡智富.基于DSP与FPGA的红外起伏背景估计系统[J].应用科技,2006.33(8)
[51]郭天天.基于FPGA的序列图像能量累加[J].信号处理,2006.22(5)
[52]王淼.红外图像实时小目标检测器的设计与实现[D].国防科学技术大学硕士学位论文.2003,12
[53]江思敏.TMS320C6000 DSP应用开发教程[M].北京:机械工业出版社,2005.2
[54]张雄伟.DSP芯片的原理与开发应用[M].北京:电子工业出版社,2006
[55]孙广富.基于管道搜索的轨迹关联检测算法及面向ASIC实现.国防科学技术大学ATR实验室文献,2006
[56]陈尚锋.基于加权动态规划的小目标检测算法研究[D].国防科学技术大学硕士学位论文,2002
[57]徐志军,徐光辉.CPLD/FPGA的开发与应用[M].北京:电子工业出版社,2002.1
[58] EDA先锋工作室,吴继华,王诚.Altera FPGA/CPLD设计(高级篇)[M].北京:人民邮电出版社,2005
[59] John G.Ackenhusen.实时信号处理——信号处理系统的设计与实现[M].北京:电子工业出版社,2002
[60]陈朝阳,孙剑伟,郑兆青等.基于多极滤波算法红外图像小目标检测的ASIC实现[J].华中科技大学学报,2005
[2]杨树谦.精确制导技术发展现状与展望[J].航天控制,2004.22(4)
[3]曾宪林.红外成像导引头及其成像制导武器评述[J].航天电子对抗,2004(5)
[4]熊辉红.外成像制导信息处理算法研究[D].国防科学技术大学博士学位论文,1998
[5]俞建成,孙胜利,陈桂林.红外图像中小目标检测研究[J].红外,2006.27(4)
[6] Zhang Bing.The predicting and matching detection algorithm of moving point target in image sequences [A] . IEEE International Conference on Neural Networks & Signal Processing[C].NanJing,China ICNNSP,2003
[7]孙德宝,周卫祥,彭家雄.红外图像序列运动小目标检测的预处理算法研究[J].红外与激光工程, 2000.29(2):12~14
[8]杨卫平,沈振康.红外图像序列小目标检测预处理技术[J].红外与激光工程,1998,2.27(1):23~28
[9]徐英.红外图像序列中运动弱小目标检测的方法研究[J].红外技术,2002,11.24(6):27~30
[10] Zang Hong,Zhao Bao jun,Mao Er-ke.A real-time effective method for infrared point target detection in spatially varying clutter[A].IEEE,CIE International Conference on radar,Proceedings [C].2001,461~464
[11] Muller M.Saliency Measures in Cluttered IR Images for ATR[J].SPIE.Apr 1999,3699: 150~154
[12] Scharfand L L,Friedlander B.Matched subspaced etectors[J].IEEE Trans.On Sig.Proc.Aug 1994,42(8):2146~2156
[13] Courtney I Hilliard.Selection of a clutter rejection algorithm for real-time target detection from an air-borne platform[A].SPIE 4048,Signal and data processing of small targets[C].2000,Orlando USA
[14] Deshpande S D, Er M H,Ronda V.Max-mean and max-median filters for detection of small targets[J].SPIE Oct 1999,3809:74~83
[15] Wang Yang ,Zhen Qinbo,Zhang Junping.Real time detection of small target in IR grey image based on mathematical morphology[J].Infrared and Laser Engineering.2003,32(1):28~31
[16] Ye Bin,Peng Jiaxiong.Small target detection based on energy accumulation andorder morphology filtering in infrared image[J].Journal of Image and Graphics,2002,7(A) (3):251~255
[17] Ye Bin,Peng Jiaxiong.Moving small target detection based on order morphology filtering in infrared image sequences [J].Journal of Data Acquisition & Processing 2001,16 (3):315~319
[18]左震,张天序.基于小波变换的红外成像海面小目标检测算法[J].红外与激光工程,1998,6.27(3):42~45
[19]朱梦宇,赵保军,韩月秋.一种实用的红外弱小目标检测跟踪处理机研究[J].激光与红外,2002,12. 32(6):407~409
[20]杨威,张田文.复杂景物环境下运动目标检测的新方法[J].计算机研究与发展,1998.35(8):724~728
[21] Rauch H. E,Futterman W. I. And Kemmer D. B.Background Suppression and Tracking with a staring Mosaic sensor[J].Optical Engineering,1981,20(1):103~110
[22] Wei P,Zeidler J,Ku W.Analysis of multiframe target detecting using pixel statistics[J].IEEE Trans,Aerosp.Electron.Syst.1995,AES-31(1):238~247
[23]倪林,俞斯乐.用于运动估计的相位相关计算[A].计算机视觉与智能控制学会第二届学术会议论文集[C].武汉,1991
[24] VegaL C,Cagigal M P,Prieto P M.Determining the velocity of a moving object from a series of photon-limited frames using a temporal Fourier transfor- m[J].SPIE.Oct 1993,1954:198~203
[25] Dong Yongqiang,Tao Ran,Zhou Siyong,et al.SAR moving target detection and imaging based on WHH Transform[J] . Journal of Beijing Institute of Technology.1999,8(1):94~100
[26] Tracie C,Moore,Stephen Lowery,Performance of Various Background Estimation and Object Detection Algorithms on Real Flight Test Interceptor Seeker Data[A].Part of the SPIE Conference on Signal and Data Processing of Small Targets [J].SPIE.July 1999,3809:86~94
[27] I. S. Reed et al.Optical Moving Target Identification Algorithm for Optical Image Sequences[J].IEEE Trans.AES,1990,26(3):434-440
[28] Thompson W B,Mutch K M,Berzins V A.Detecting moving objects using the rigidity constraint[J] . IEEE Transactions on Pattern Analysis and Machine Intelligence.1993,15:162~166
[29] Nishiguchi K,Kobayashi M,Ichikawa A.Small target detection from imagesequences using recursive max filter [J].SPIE.1995,2561:153~165
[30] Diani M,Corsiniq Baldacci.A Space-time processing for the detection of air borne targets in IR image sequences[J].IEE Proceeding of Vision,Image and Signal Processing.Jun 2001,148(3):151~157
[31]彭嘉雄,彭铁.弱目标检测的图像流法[J].红外与激光工程,1996,8,25(4):34~40
[32]李国宽,李红,彭嘉雄.从图像序列中检测定位小目标的一种方法[J].华中理工大学学报.1999,8,27(8):63~65
[33] Cheung J F,Y Wicks M C,Genello G J.A Statistical Theory for Optimal Detection of Moving Objects in Variable Corruptive Noise[J].IEEE Transactions on Image Process- ing.Dec 1999,8(12):1772~1787
[34] Choi J,Jang J,Lee S,et.al.Multiple moving objecte stimation in images equences of natural scene[J].Opt Eng.1997,36(8):2176~2183
[35] Blostein S,D Richardson H S.A sequential detection approach to target tracking[J].IEEE Trans.Aerospace and Electronic Systems.1994,30(1):197~211
[36] Blostein S D,Huang T S.Detection small moving objects in image sequences using sequential hypothesis testing[J].IEEE Trans Signal Processing.1991,39(7):611~1629
[37] Collet C,Fablet R.Motion Estimation on Infrared Image Sequences[J].SPIE.1999 3699:67~77
[38] Thomas Fechner,Rainer Hach,Oliver Rockinger,Christoph Stahl.Detection and tracking of small target in aerial image sequences under unknown sensor motion[A].Part of the SPIE Conference on Signal Processing,Sensor Fusion,and Target Recognition VII[C],Orlando,Florida,April 1998,SPIE 3374:259~266
[39] Askar H,LI Xiao feng,LI Zai ming.Background clutter suppression and dim moving point targets detection using nonparametric method[A].IEEE 2002 International Conference on Communications,Circuits and Systems and West Sino Expositions [C].2002:982~986
[40] Grossman S B,Emmons R B.Performance analysis and size optimization of focal planes for point-source tracking algorithm applications[J].Optical Engineer- ing.Apr 1984,23(2):167~176
[41]熊艳,彭嘉雄,丁明跃等.一种低信噪比条件下点目标检测方法[J].电子学报.1996,12. 24(12):41~45
[42]张金锁,宋利权,靳民会等.基于带通滤波的红外小目标捕获算法研究[J].红外与激光工程,2002,8.31(4):294~296
[43] Lion R,Inigom,Azimi-Sadjad E S.Dim target detection using high order correlation method[J].IEEE Trans Aerosp,Electron,Syst.1994,AES-29(1):197~212
[44]黄林梅,基于动态规划的红外运动小目标检测算法[J].红外激光与工程,2004.33(3)
[45] Chen HuaMing,Sun Guangfu,Lu Huanzhang,Chang Qing.Multi-level thresholding and confidence testing detection algorithm for small moving target in image sequences[J].SPIE,Vol. 4473,2001:520~529
[46]周冰,王永仲,应家驹.弱小目标检测技术浅析[J].红外技术,2007.29(1)
[47]陈玉丹,周冰.红外小目标检测中的背景预处理技术研究[J].科学技术与工程,2006.6(18)
[48]李学广.红外图像预处理算法研究及硬件实现[D].哈尔滨工程大学硕士学位论文,2007
[49]方强,孙广富,卢焕章.红外焦平面图像稳像处理的FPGA实现[J].微处理机, 2001(1)
[50]蔡智富.基于DSP与FPGA的红外起伏背景估计系统[J].应用科技,2006.33(8)
[51]郭天天.基于FPGA的序列图像能量累加[J].信号处理,2006.22(5)
[52]王淼.红外图像实时小目标检测器的设计与实现[D].国防科学技术大学硕士学位论文.2003,12
[53]江思敏.TMS320C6000 DSP应用开发教程[M].北京:机械工业出版社,2005.2
[54]张雄伟.DSP芯片的原理与开发应用[M].北京:电子工业出版社,2006
[55]孙广富.基于管道搜索的轨迹关联检测算法及面向ASIC实现.国防科学技术大学ATR实验室文献,2006
[56]陈尚锋.基于加权动态规划的小目标检测算法研究[D].国防科学技术大学硕士学位论文,2002
[57]徐志军,徐光辉.CPLD/FPGA的开发与应用[M].北京:电子工业出版社,2002.1
[58] EDA先锋工作室,吴继华,王诚.Altera FPGA/CPLD设计(高级篇)[M].北京:人民邮电出版社,2005
[59] John G.Ackenhusen.实时信号处理——信号处理系统的设计与实现[M].北京:电子工业出版社,2002
[60]陈朝阳,孙剑伟,郑兆青等.基于多极滤波算法红外图像小目标检测的ASIC实现[J].华中科技大学学报,2005