静态背景下的点源动目标探测算法研究
|
摘要
基于图像信息的运动目标检测技术,在军事及工业领域均获得广泛的应用。当光学成像系统和目标的相对位置较远时,目标在图像平面上表现为点状或不稳定斑点状,即点目标。运动点目标检测技术,是信号和图像处理领域的热点研究课题。该项技术的实现,能够有效的扩大光学成像系统的实际作用距离,使得基于图像信息对目标进行远程监控及目标的及时捕捉成为可能,具有重要的应用价值。本文主要针对处于静态背景下的运动点目标探测问题进行了深入的探讨和研究,所完成的工作和成果主要有以下四个方面:
1.分析了点目标检测的技术难点,研究了点目标检测技术的发展状况,并对现有的具有代表性的点目标检测方法进行简单分析,包括算法的理论基础、优缺点和适用范围,并在此基础上,总结点目标检测算法的发展方向:简单,计算量少,可靠性高和易于硬件实现。
2.针对视频图像背景可能存在的抖动和缓慢的平移运动,讨论了一种基于灰度投影的背景运动校正算法,通过灰度投影相关运算对背景的平移运动矢量进行估计,实现背景平移偏差的校正,算法利用图像的纹理信息实现投影分区的自动筛选,有效剔除了灰度信息含量少、对背景运动不敏感的投影分区,减少了不必要的计算,提高了背景运动矢量估计的准确度,实验结果证明了算法的有效性。
3.研究了点目标图像预处理技术,包括背景抑制和目标分割技术。重点分析了基于空域高通滤波的背景抑制算法和基于数学形态学的背景抑制算法。目标分割采用阈值法,在对几种常用阈值分割算法分析的基础上,采用一种改进的局部自适应阈值法对点目标图像进行分割。
4.根据点目标运动的连续性,提出一种基于图像序列的运动点目标检测算法,该算法采用质心点位置代替目标点位置进行检测,通过邻域搜索判决、多帧航迹检测置信度判决和目标航迹确认,实现点目标轨迹的提取,在仿真实验中取得了很好的结果。
The moving target detection technology based on image information, has been widely used in military and civil fields. When there is a far distance between the optical imaging system and moving objects, the image of moving objects appears as the shape of point or spots, which is the point target. The moving point target detection technology has been a hot issue in the area of signal and image processing. The realization of this technology could effectively expand the working range of optical-sensors, so as to make it possible to monitor and capture the moving point targets in real time, which has great application value. In the paper the technology of moving point target detection of static background is deeply studied in detail. Main research works and achievements of this thesis are listed as follows:
1. The technical difficulties of moving point target detection is analyzed. The development situations of this technology is widely studied and the current representative methods of it are studied, including theoretical foundations, advantages and disadvantages and the sphere of application. Based on above work, the research direction of moving point target detection technology is summarized: simplicity, less time consuming, high reliability and the easy realization on hardware.
2. Focusing on the undesirable influence of background jitter and slow translational motion of video images, one grey projection algorithm is discussed in this paper, taking advantage of grey projection correlative operation to estimate the translational motion vector so as to realize the adjustment of the translation error. The algorithm uses the image texture information to implement the automatic selection in different projection sub areas which could effectively removed the projection area with less grey information content and insensitivity to the movement. This method reduces the unnecessary calculation and avoids the estimated error to motion vector due to inaccurate estimates.
3. The technology of image pre-processing is studied, including background suppression algorithms and target segmentation algorithms. Two methods of pre-processing of image are studied in details, which are the algorithm based on high-pass filtering theory and the algorithm based on mathematical morphology theory. On the basis of analyzing about some representative threshold division methods, the self-adapting threshold division method is used to extract suspicious points.
4. According to the continuity of the point target, a moving target detection algorithm based on image sequences is proposed. This algorithm adopts the center of mass instead of target points to detect the location, and extracts the track of point target through neighborhood search judgment, multi-frame track detection confidence judgment and target’s track confirmation. The simulation result shows that, this algorithm can effectively detect point target in static background.
1.分析了点目标检测的技术难点,研究了点目标检测技术的发展状况,并对现有的具有代表性的点目标检测方法进行简单分析,包括算法的理论基础、优缺点和适用范围,并在此基础上,总结点目标检测算法的发展方向:简单,计算量少,可靠性高和易于硬件实现。
2.针对视频图像背景可能存在的抖动和缓慢的平移运动,讨论了一种基于灰度投影的背景运动校正算法,通过灰度投影相关运算对背景的平移运动矢量进行估计,实现背景平移偏差的校正,算法利用图像的纹理信息实现投影分区的自动筛选,有效剔除了灰度信息含量少、对背景运动不敏感的投影分区,减少了不必要的计算,提高了背景运动矢量估计的准确度,实验结果证明了算法的有效性。
3.研究了点目标图像预处理技术,包括背景抑制和目标分割技术。重点分析了基于空域高通滤波的背景抑制算法和基于数学形态学的背景抑制算法。目标分割采用阈值法,在对几种常用阈值分割算法分析的基础上,采用一种改进的局部自适应阈值法对点目标图像进行分割。
4.根据点目标运动的连续性,提出一种基于图像序列的运动点目标检测算法,该算法采用质心点位置代替目标点位置进行检测,通过邻域搜索判决、多帧航迹检测置信度判决和目标航迹确认,实现点目标轨迹的提取,在仿真实验中取得了很好的结果。
The moving target detection technology based on image information, has been widely used in military and civil fields. When there is a far distance between the optical imaging system and moving objects, the image of moving objects appears as the shape of point or spots, which is the point target. The moving point target detection technology has been a hot issue in the area of signal and image processing. The realization of this technology could effectively expand the working range of optical-sensors, so as to make it possible to monitor and capture the moving point targets in real time, which has great application value. In the paper the technology of moving point target detection of static background is deeply studied in detail. Main research works and achievements of this thesis are listed as follows:
1. The technical difficulties of moving point target detection is analyzed. The development situations of this technology is widely studied and the current representative methods of it are studied, including theoretical foundations, advantages and disadvantages and the sphere of application. Based on above work, the research direction of moving point target detection technology is summarized: simplicity, less time consuming, high reliability and the easy realization on hardware.
2. Focusing on the undesirable influence of background jitter and slow translational motion of video images, one grey projection algorithm is discussed in this paper, taking advantage of grey projection correlative operation to estimate the translational motion vector so as to realize the adjustment of the translation error. The algorithm uses the image texture information to implement the automatic selection in different projection sub areas which could effectively removed the projection area with less grey information content and insensitivity to the movement. This method reduces the unnecessary calculation and avoids the estimated error to motion vector due to inaccurate estimates.
3. The technology of image pre-processing is studied, including background suppression algorithms and target segmentation algorithms. Two methods of pre-processing of image are studied in details, which are the algorithm based on high-pass filtering theory and the algorithm based on mathematical morphology theory. On the basis of analyzing about some representative threshold division methods, the self-adapting threshold division method is used to extract suspicious points.
4. According to the continuity of the point target, a moving target detection algorithm based on image sequences is proposed. This algorithm adopts the center of mass instead of target points to detect the location, and extracts the track of point target through neighborhood search judgment, multi-frame track detection confidence judgment and target’s track confirmation. The simulation result shows that, this algorithm can effectively detect point target in static background.
引文
[1]许彬,郑链,王永学等,红外序列图像小目标检测与跟踪技术综述,红外与激光工程,2004,33(5):482～487
[2]张惠娟,梁彦,程咏梅等,运动弱小目标先跟踪后检测技术的研究进展,红外技术,2006,28(7):423～429
[3]王伞,红外弱小目标检测技术研究:[硕士论文],哈尔滨;哈尔滨工程大学,2005
[4] G.A.Lampropoulos,J.F.Boulter,Filtering of Moving Targets Using SBIR Sequential Frames, IEEE Trans On Aerospace and Electronic System,1995,31(4):1255～1266
[5] Reed,I. S. Gagliardi,Optical Moving Target Detection with 3-D Matched filtering,IEEE Trans. on Aerospace and Electronic System,1988,24(4):327～336
[6] S. D. Blostein,T. S. Huang,Detecting small moving objects in image sequences using sequential hypothesis testing,IEEE Trans. on Signal Processing,1991,39 (7):1611～1629
[7] Stocker A D,Jensen P,Algorithms and architectures for implementing large velocity filter banks.SPIE,1991,1481:l40～155
[8]李斌,彭嘉雄,基于动态规划的红外小目标检测与识别,华中理工大学学报,2000,28(6):68～70
[9]黄林梅,张桂林,王新余,基于动态规划的红外运动小目标检测算法,红外与激光工程,2004,33(3):303～319
[10]钟圣芳,张兵,卢焕章,一种基于动态规划的点目标轨迹关联算法,计算机测量与控制,2004,12(8):772～775
[11]陈尚锋,陈华明,卢焕章,基于加权动态规划和航迹关联的小目标检测技术,国防科技大学学报,2003,25(2):46～50
[12] LiouR,Azimi-Sadjadi M R Dent R,Dim Target detection using high Order correlation method,IEEE Transactions on Aerospace and Electronic Systems,1993,29(3):841～856.
[13]陈非,敬忠良,李建勋,红外与毫米波学报,红外序列图像中缓动点目标的投影检测算法及其改进,2003,22(2):96～100
[14] G. H. Watson,S. K. Watson,The Detection of Moving Targets in Time-sequenced Imagery Using Statistical Background Rejection,SPIE, 3163,45～53
[15]吴荣华,王江安,宗思光等,红外技术,基于形态滤波与梯度检测的红外弱点目标检测技术研究,2007,29(4):234～238
[16]温佩芝,史泽林,于海斌,光电工程,基于形态学的海面背景红外点目标检测方法,2003,30(6):55～58
[17]张文超,岩飞,陈贺新,基于Tophat变换的复杂背景下运动点目标识别算法,中国图象图形学报,2007,12(5):871～874
[18]刘志军,郑兆青,桑红石,基于跳跃马尔可夫模型的红外点目标检测方法,华中科技大学学报(自然科学版),2007,35(1):23～25
[19]罗中良,李昕,基于小波变换的红外弱小目标检测新方法,红外技术,2006,28(8):456～459
[20]党莹,翟林培,基于小波多分辨力分析的低信杂比下点目标跟踪方法,光电工程,2002,27(5):47～51
[21]李红艳,图像的信噪比小目标检测与跟踪算法研究:[博士论文],西安;西安电子科技大学,2000
[22]石佼,李晋惠,电子稳像技术算法研究,理论与方法,2007,26(3): 18～24
[23]安晓东,山西电子技术,电子稳像原理及方法概述,2006,6:83~84
[24]石佼,李晋惠,理论与方法,电子稳像技术算法研究,2007,26(3):18~24
[25]赵红颖,金宏,熊经武,光学精密工程,电子稳像技术概述,2001,9(4):353~359
[26]钟平,于前洋,金光等,机载摄像设备图像稳定方法探讨,光电工程,2002,29:73～76
[27]陈志坚,李哲,石磊,科学技术与工程,一种基于图像特征块匹配的电子稳像算法块匹配,2007,11(7):2512~2515
[28]赵芹,周涛,舒勤,基于特征点的图像配准技术探讨,红外技术,2006,28(6):327~330
[29]张二虎,卞正中,基于最大熵和互信息最大化的特征点配准算法,计算机研究与发展,2004,41(7):1194~1199
[30] Taejung. Kim,Yong-Jo Im,Automatic satellite image registration by combination of matching and random sample consensus,Geoscience and Remote Sensing, IEEE Transactions on,2003,41(5):1111~1117
[31] McGuire. M,Stone. H. S,Techniques for multiresolution image registration in the presence of occlusions,Geoscience and Remote Sensing, IEEE Transactions on,2000,38(3):1476~1479
[32]孙辉,快速灰度投影算法及其在电子稳像中的应用,光学精密工程,2007,15(3):412～416
[33]孙远,周刚慧,赵立初等,上海交通大学学报,灰度图像匹配的快速算法,2000,34(5):702~704
[34]朱娟娟,郭宝龙,冯宗哲,光子学报,一种基于灰度投影算法的电子稳像方法,2005,34(8):1266~1269
[35]周渝斌,赵跃进,北京理工大学学报,基于单向投影矢量的数字电子稳像方法,2003,23(4):509~512
[36]刘江一,全景成像技术若干关键问题的研究:[硕士论文],天津;天津大学,2007
[37]张竞,动态图像序列中运动目标检测若干技术问题的研究:[硕士论文],天津;天津大学,2007
[38]陈颖,序列图像中微弱点状运动目标检测及跟踪技术研究:[博士论文],成都;电子科技大学,2002
[39]张孝霖等,在海域红外杂波背景中小目标探测的空间匹配滤波技术,红外与毫米波学报,1994,13(2):95～101.
[40]夏爱利,马彩文,张砺佳,一种基于高通滤波的背景抑制方法,科学技术工程,2007, 12(7):2978～2980
[41]叶斌,基于顺序形态滤波的运动点目标检测,红外与激光工程,2001, 30(1):25～29
[42]张宇,吴宏刚,陈跃斌等,采用形态神经网络背景自适应预测的图像弱小目标检测,计算机应用研究,2007,24(3):289～291
[43]林长春,孙胜利,陈桂林,复杂背景下含点目标单帧红外图像的预处理,激光与红外,2006,36(9):900～903
[44]王耀南,李树涛,毛建旭,计算机图像处理与识别技术,北京:高等教育出社,2001:83～89
[45]刘玲,红外弱小目标检测算法研究:[硕士论文],哈尔滨;哈尔滨工程大学,2005
[46] (美)冈萨雷斯,数字图像处理(MATLAB版):(阮秋琦等译):北京:电子工业出版社,2005:252～284
[47]王岩,动态背景下目标跟踪若干技术问题研究:[硕士论文],天津;天津大学,2007
[48]高智,吴巍,红外小目标的背景预测与分割,武汉理工大学学报,2006,28(11),178～182
[49]周铭,低信噪比红外小目标图像分割方法,激光与红外,2004,34(3):225~228
[50]王俊芳,基于形态学的微弱运动目标检测技术及应用:[硕士论文],成都;电子科技大学,2007
[51]张海英,张田文,一种新的用于点目标检测与跟踪的闭环pipeline,计算机学报,2005,28(3):378～383
[2]张惠娟,梁彦,程咏梅等,运动弱小目标先跟踪后检测技术的研究进展,红外技术,2006,28(7):423～429
[3]王伞,红外弱小目标检测技术研究:[硕士论文],哈尔滨;哈尔滨工程大学,2005
[4] G.A.Lampropoulos,J.F.Boulter,Filtering of Moving Targets Using SBIR Sequential Frames, IEEE Trans On Aerospace and Electronic System,1995,31(4):1255～1266
[5] Reed,I. S. Gagliardi,Optical Moving Target Detection with 3-D Matched filtering,IEEE Trans. on Aerospace and Electronic System,1988,24(4):327～336
[6] S. D. Blostein,T. S. Huang,Detecting small moving objects in image sequences using sequential hypothesis testing,IEEE Trans. on Signal Processing,1991,39 (7):1611～1629
[7] Stocker A D,Jensen P,Algorithms and architectures for implementing large velocity filter banks.SPIE,1991,1481:l40～155
[8]李斌,彭嘉雄,基于动态规划的红外小目标检测与识别,华中理工大学学报,2000,28(6):68～70
[9]黄林梅,张桂林,王新余,基于动态规划的红外运动小目标检测算法,红外与激光工程,2004,33(3):303～319
[10]钟圣芳,张兵,卢焕章,一种基于动态规划的点目标轨迹关联算法,计算机测量与控制,2004,12(8):772～775
[11]陈尚锋,陈华明,卢焕章,基于加权动态规划和航迹关联的小目标检测技术,国防科技大学学报,2003,25(2):46～50
[12] LiouR,Azimi-Sadjadi M R Dent R,Dim Target detection using high Order correlation method,IEEE Transactions on Aerospace and Electronic Systems,1993,29(3):841～856.
[13]陈非,敬忠良,李建勋,红外与毫米波学报,红外序列图像中缓动点目标的投影检测算法及其改进,2003,22(2):96～100
[14] G. H. Watson,S. K. Watson,The Detection of Moving Targets in Time-sequenced Imagery Using Statistical Background Rejection,SPIE, 3163,45～53
[15]吴荣华,王江安,宗思光等,红外技术,基于形态滤波与梯度检测的红外弱点目标检测技术研究,2007,29(4):234～238
[16]温佩芝,史泽林,于海斌,光电工程,基于形态学的海面背景红外点目标检测方法,2003,30(6):55～58
[17]张文超,岩飞,陈贺新,基于Tophat变换的复杂背景下运动点目标识别算法,中国图象图形学报,2007,12(5):871～874
[18]刘志军,郑兆青,桑红石,基于跳跃马尔可夫模型的红外点目标检测方法,华中科技大学学报(自然科学版),2007,35(1):23～25
[19]罗中良,李昕,基于小波变换的红外弱小目标检测新方法,红外技术,2006,28(8):456～459
[20]党莹,翟林培,基于小波多分辨力分析的低信杂比下点目标跟踪方法,光电工程,2002,27(5):47～51
[21]李红艳,图像的信噪比小目标检测与跟踪算法研究:[博士论文],西安;西安电子科技大学,2000
[22]石佼,李晋惠,电子稳像技术算法研究,理论与方法,2007,26(3): 18～24
[23]安晓东,山西电子技术,电子稳像原理及方法概述,2006,6:83~84
[24]石佼,李晋惠,理论与方法,电子稳像技术算法研究,2007,26(3):18~24
[25]赵红颖,金宏,熊经武,光学精密工程,电子稳像技术概述,2001,9(4):353~359
[26]钟平,于前洋,金光等,机载摄像设备图像稳定方法探讨,光电工程,2002,29:73～76
[27]陈志坚,李哲,石磊,科学技术与工程,一种基于图像特征块匹配的电子稳像算法块匹配,2007,11(7):2512~2515
[28]赵芹,周涛,舒勤,基于特征点的图像配准技术探讨,红外技术,2006,28(6):327~330
[29]张二虎,卞正中,基于最大熵和互信息最大化的特征点配准算法,计算机研究与发展,2004,41(7):1194~1199
[30] Taejung. Kim,Yong-Jo Im,Automatic satellite image registration by combination of matching and random sample consensus,Geoscience and Remote Sensing, IEEE Transactions on,2003,41(5):1111~1117
[31] McGuire. M,Stone. H. S,Techniques for multiresolution image registration in the presence of occlusions,Geoscience and Remote Sensing, IEEE Transactions on,2000,38(3):1476~1479
[32]孙辉,快速灰度投影算法及其在电子稳像中的应用,光学精密工程,2007,15(3):412～416
[33]孙远,周刚慧,赵立初等,上海交通大学学报,灰度图像匹配的快速算法,2000,34(5):702~704
[34]朱娟娟,郭宝龙,冯宗哲,光子学报,一种基于灰度投影算法的电子稳像方法,2005,34(8):1266~1269
[35]周渝斌,赵跃进,北京理工大学学报,基于单向投影矢量的数字电子稳像方法,2003,23(4):509~512
[36]刘江一,全景成像技术若干关键问题的研究:[硕士论文],天津;天津大学,2007
[37]张竞,动态图像序列中运动目标检测若干技术问题的研究:[硕士论文],天津;天津大学,2007
[38]陈颖,序列图像中微弱点状运动目标检测及跟踪技术研究:[博士论文],成都;电子科技大学,2002
[39]张孝霖等,在海域红外杂波背景中小目标探测的空间匹配滤波技术,红外与毫米波学报,1994,13(2):95～101.
[40]夏爱利,马彩文,张砺佳,一种基于高通滤波的背景抑制方法,科学技术工程,2007, 12(7):2978～2980
[41]叶斌,基于顺序形态滤波的运动点目标检测,红外与激光工程,2001, 30(1):25～29
[42]张宇,吴宏刚,陈跃斌等,采用形态神经网络背景自适应预测的图像弱小目标检测,计算机应用研究,2007,24(3):289～291
[43]林长春,孙胜利,陈桂林,复杂背景下含点目标单帧红外图像的预处理,激光与红外,2006,36(9):900～903
[44]王耀南,李树涛,毛建旭,计算机图像处理与识别技术,北京:高等教育出社,2001:83～89
[45]刘玲,红外弱小目标检测算法研究:[硕士论文],哈尔滨;哈尔滨工程大学,2005
[46] (美)冈萨雷斯,数字图像处理(MATLAB版):(阮秋琦等译):北京:电子工业出版社,2005:252～284
[47]王岩,动态背景下目标跟踪若干技术问题研究:[硕士论文],天津;天津大学,2007
[48]高智,吴巍,红外小目标的背景预测与分割,武汉理工大学学报,2006,28(11),178～182
[49]周铭,低信噪比红外小目标图像分割方法,激光与红外,2004,34(3):225~228
[50]王俊芳,基于形态学的微弱运动目标检测技术及应用:[硕士论文],成都;电子科技大学,2007
[51]张海英,张田文,一种新的用于点目标检测与跟踪的闭环pipeline,计算机学报,2005,28(3):378～383