双波段红外图像目标特性分析及其融合方法研究
|
摘要
图像融合技术弥补了单一传感器的不足,具有大幅提高系统的探测性能、提高系统的自动化程度、降低系统的整体成本、提高探测识别的客观性和正确性等优点,因此图像融合成为现在一个研究的热点,对于双波段图像融合来讲为了达到比较好的融合效果,首先是要分析目标辐射特性、辐射传输特性、成像特性和图像特征等在不同波段的异同,在此基础上研究图像融合算法,利用互补信息和冗余信息来增强图像融合效果,才能取得好的融合效果,使融合具有针对性。
本文首先主要研究了目标特性分析中所需要的红外辐射基本理论,在红外辐射理论研究的基础上,研究了目标背景辐射传输特性,对目标背景辐射对比度进行了计算,分析了不同波段目标背景辐射对比度的特点以及由此所形成的图像特征,并结合具体的红外图像采用直方图、均值、标准偏差、局部熵、调和熵、多向粗糙度等对不同波段的图像细节信息进行了具体阐述,最后结合目标背景辐射传输特性,提出了基于普朗克定律的双波段红外图像融合算法,结合不同波段的图像特征,提出了基于目标特征的双波段红外图像融合算法。通过编程实现,取得了比较好的融合结果,图像具有比较丰富的细节信息,目标清晰,能够比较全面的反映出了目标的整体特征,使目标更容易被识别和探测,也使融合具有了针对性。
The technique of image fusion makes up the lack of single sensor, it has the virtue of improving performance of systemic detection、improving roboticized degree、reducing the systemic cost、improving objectivity and validity of detection identifying ,so image fusion is the hotspot of research now. If the people want to achieve the aim of the better effect of fusion, first, they must analyze the difference of feature of radiation of target、the difference of feature of transmission of radiation、the difference of feature of imaging and the difference of character of image in the different bands , on base of the analysis the fusion arithmetic is researched, the effect of image fusion is toned by using the commutative information and the redundancy information, the people can achieve the good effect of fusion and the fusion has the pertinence on base of target feature and research of fusion arithmetic are linked.
The paper researches the theory that the analyzing of feature of target need of infrared radiation, then the paper researches the feature of radiation transmission and compute the radiation contrast of target and background, at the same time the paper analyzes the feature of the radiation contrast of target and background and the image feature that the radiation contrast of target and background arose, then the paper expounds the image feature using he bar chart、mean、standard deviation、local entropy、congruity of entropy、roughness concentration of multi direction, at last the paper bring forward the fusion arithmetic of double bands infrared on base of fundamentals of Plank and bring forward the fusion arithmetic double bands infrared on base of feature of target. the paper realizes the fusion arithmetic using MATLAB, the effect is better, the fusion image has the abundant detail information , the target is clarity, the feature of target is generally reflected, the target is identified and detected in easy. Besides the fusion arithmetic has the pertinence.
本文首先主要研究了目标特性分析中所需要的红外辐射基本理论,在红外辐射理论研究的基础上,研究了目标背景辐射传输特性,对目标背景辐射对比度进行了计算,分析了不同波段目标背景辐射对比度的特点以及由此所形成的图像特征,并结合具体的红外图像采用直方图、均值、标准偏差、局部熵、调和熵、多向粗糙度等对不同波段的图像细节信息进行了具体阐述,最后结合目标背景辐射传输特性,提出了基于普朗克定律的双波段红外图像融合算法,结合不同波段的图像特征,提出了基于目标特征的双波段红外图像融合算法。通过编程实现,取得了比较好的融合结果,图像具有比较丰富的细节信息,目标清晰,能够比较全面的反映出了目标的整体特征,使目标更容易被识别和探测,也使融合具有了针对性。
The technique of image fusion makes up the lack of single sensor, it has the virtue of improving performance of systemic detection、improving roboticized degree、reducing the systemic cost、improving objectivity and validity of detection identifying ,so image fusion is the hotspot of research now. If the people want to achieve the aim of the better effect of fusion, first, they must analyze the difference of feature of radiation of target、the difference of feature of transmission of radiation、the difference of feature of imaging and the difference of character of image in the different bands , on base of the analysis the fusion arithmetic is researched, the effect of image fusion is toned by using the commutative information and the redundancy information, the people can achieve the good effect of fusion and the fusion has the pertinence on base of target feature and research of fusion arithmetic are linked.
The paper researches the theory that the analyzing of feature of target need of infrared radiation, then the paper researches the feature of radiation transmission and compute the radiation contrast of target and background, at the same time the paper analyzes the feature of the radiation contrast of target and background and the image feature that the radiation contrast of target and background arose, then the paper expounds the image feature using he bar chart、mean、standard deviation、local entropy、congruity of entropy、roughness concentration of multi direction, at last the paper bring forward the fusion arithmetic of double bands infrared on base of fundamentals of Plank and bring forward the fusion arithmetic double bands infrared on base of feature of target. the paper realizes the fusion arithmetic using MATLAB, the effect is better, the fusion image has the abundant detail information , the target is clarity, the feature of target is generally reflected, the target is identified and detected in easy. Besides the fusion arithmetic has the pertinence.
引文
[1] 李敏, 谈正, 李晓燕. 多传感器图像融合技术及评价方法研究[J]. 计算机工程与应用, 2006, 28(6): 39~41.
[2] 侯俊, 苗壮, 潘泉. 基于DSmT的序列图像智能融合目标识别方法[J]. 计算机应用. 2006, 26(1): 120~122.
[3] A. Roblin, I. Dubois, F. Grisch et al. Comparison Between Computations and Measurements of a H2/LOX Rocket Motor Plume[C]. 8th AIAA/ASME Joint Thermophysics and heat transfer Conference. AIAA-2002-307:1~10.
[4] Filip Neele, Ric Schleijpen. Electro-Optical Missile Plume Detection[C]. Targets and backgrounds IX: Characterization and Representation. Proceedings of SPIE. Vol 5075, 2003: 270~280.
[5] 宣益民, 韩玉阁. 地面目标与背景的红外特征[M]. 北京: 国防工业出版社, 2004.
[6] 宋自林, 杜益, 黄松. 从“沙漠之狐”行动与科索沃战争看传感器技术与数据融合[J]. 军事电子信息系统专题技术文选. 电子部 28 所. 2000. l3 (1).
[7] 艾雨兵, 郑金华. 舰船电子装备特点及其发展[J]. 情报指挥控制系统与仿真技术, 2004, 16(1): 38~42.
[8] Cuthbertson G M, Seedhouse A. The evolution of a digital dual waveband FLIR[C]. SPIE. 1994, 2269: 665~677.
[9] Clark G A, et al. Land mine detection using multispectral image fusion[C]. Proceedings of the Symposium on Autonomous Vehicles in Mine Countermeasures, Monterey, CA(United States), 1995, 3~7.
[10] 范之国, 高俊, 胡良梅等. 夜视图像自适应彩色化融合算法[J]. 仪器仪表学报. 2006, 27(6): 733~737.
[11] 倪国强, 戴文, 李勇量, 蒲恬. 基于响尾蛇双模式细胞机理的可见光/红外图像彩色融合技术的优势和前景展望[J]. 北京理工大学学报. 2004, 24(2):95~100.
[12] 美国国防部关键技术计划 (1992 财年). 北京: 国防科技信息中心, 1993.
[13] 王宏, 敬忠良, 李建勋. 多分辨率图像融合的研究与发展[J]. 控制理论与应用. 2004, 21(1): 145~151.
[14] 高稚允, 金伟其等. 一种可实时化的多光谱图像融合系统[J]. 光学技术. 1995,4:13~16.
[15] 王毅, 倪国强, 李勇量. 基于 TMS320C6x 的双波段图像高速融合系统[J]. 2002. 7(A):10: 1038~1042.
[16] 倪国强. 多波段图像融合算法研究及其新发展. 光电子技术与信息[J]. 2001,14(5): 11~17.
[17] Peter J B, Edward H A. Merging image through pattern decomposition[C]. SPIE.1985, 575: 173~181.
[18] Toet A, Lodewik J, Ruyven V, et al.. Merging thermal and visual image by a contrast pyramid[J]. Optical Engineering. 1992, 28: 789~792.
[19] Steinnocher K. Application of adaptive filters for multisensoral image fusion[J]. Geoscience and Remote Sensing, 1997, 2: 910~912.
[20] Tery A.Wilson, Steven K.Rogers, Lemuel R.Myers. Perceptual-based hyperspectral image fusion using multiresolution analysis[J]. Optical Engineering. 1995, 34: 3154~3164.
[21] Kundur D, Hatzinakos D. A novel approach to mulitispectral blind image fusion[C]. SPIE. 1997, 3067: 83~93.
[22] Smart S, Scarff L A. Combining visual and IR images for sensor fusion —two approaches[C]. SPIE. 1992, 1668: 102~112.
[23] Yocky D A. Artifacts in Wavelet Image Merging[C].SPIE. 1996, 35(7): 2094~2101.
[24] Ajjimarange P. Neural network model for fusion fo visible and infrared sensor outputs[C]. SPIE. 1988, 1003: 153~160.
[25] Kinser J M. Pulse-coupled image fusion[J]. Optical Engineering. 1997, 36(3): 737~742.
[26] Duane G. Pixel-level sensor fusion for improved object recognition[C]. SPIE. 1998, 931: 180~185.
[27] Streit R L, Luginbuhl T E. Maximum likelihood training of probabilistic neural networks[J]. IEEE Trans. Neur. Networks, 1994 , 5: 764~783.
[28]Peter J B, Edward H A. Merging image through pattern decomposition[C]. SPIE. 1985, 575: 173~181.
[29] 张建奇. 红外物理[M].西安: 西安电子科技大学出版社, 2004: 105~114.
[30] 张敬贤, 李玉丹, 金伟其. 微光与红外成像技术[M]. 北京: 北京理工大学出版社, 1995.
[31] 张智丰, 李向新, 彭群生. 基于大气传输模型的动态目标红外成像仿真[J]. 系统仿真学报.2000,12(9):524-527
[32]王毅, 范伟, 饶瑞中. 典型大气条件下的目标-背景对比度的计算分析[J]. 激光与红外. 2004, 34(5):375~378.
[33]范伟, 王毅, 饶瑞中. 根据大气辐射特征进行目标探测的波段选择[J]. 红外与激光工程. 2002, 34(2):177~182.
[34] 邢强林, 黄惠明, 熊仁生,于 涛. 红外成像探测系统作用距离分析方法研究.光子学报[J].2004,33(4): 894-896.
[35] 王江安, 肖伟岸, 申林. 海空背景下目标红外辐射特征分析.海军工程大学学报[J].2001,13(4): 29-31
[36]吴文健, 胡碧茹, 满亚辉译. 地面目标和背景的热红外特性[M].国防工业出版社.2004
[37]吴北婴, 李卫, 陈洪滨等. 大气传输实用算法[M]. 北京: 气象出版社, 1998.
[38] 吴晗平. 红外辐射大气透过率的工程理论计算方法研究。光学精密工程[J]. 1998,8(4):38-43
[39] 路 远, 凌永顺. 红外辐射大气透射比的简易计算.红外技术[J].2003,25(5): 45-49
[40] 王春勇, 金伟其. 光电成像系统分析中大气传输特性计算的几个问题. 北京理工大学学报[J].2003,23(5): 617-620
[41] Pieter A. Jacobs. 吴文健等译.地面目标和背景的热红外特性[M]. 北京: 国防工业出版社. 2004.
[42] 王江安, 肖伟岸. 基于双波段的目标红外辐射特征分析[J]. 激光与红外. 2001, 31(6):351~353.
[43]金剑秋, 王章野, 彭群生. 相关准则系数下的多光谱图像最优融合方法.计算机辅助设计与图形学学报[J]. 1~6
[44] A.Goldberg, T.Fischer, S.Kennerly. Dual-Band Imaging of Military Targets Using a QWIP Focal Plane Array. EO/IR Technology Branch, Army Research Laboratory, Adelphi, MD 20783
[45] Michelle Tomkinson, Brian Teaney, Jeffrey Olson. Dual Band Sensor Fusion for Urban Target Acquisition. U.S. Army RDECOM CERDEC Night Vision and ElectronicSensors Directorate10221 Burbeck RdFt. Belvoir, Virginia 22060-5806
[46] 汪 强, 刘钢钦. 一种新的基于局部小波统计特性的多聚焦图像融合算法.长沙大学学报[J].2006,20(6): 72-74
[47] 张焕龙, 杨国胜, 王婷,申晓华. 基于局部熵和渐变因子的自适应图像拼接.计算机工程与应用[J].2006,22(3): 72-74
[48] 王江安, 朱向前, 宗思光,肖伟岸. 红外目标特征分析及融合特征提取.传感技术学报[J].2005,6(2): 289-291
[49] 林国余, 张为公. 基于进化规划的最大类间方差的图像分割算法.传感技术学报[J].2006,2(1): 179-182
[50] 李寒松, 鲁传运. 红外舰船小目标的检测. 红外与激光工程[J].2006,35(4): 495-498
[2] 侯俊, 苗壮, 潘泉. 基于DSmT的序列图像智能融合目标识别方法[J]. 计算机应用. 2006, 26(1): 120~122.
[3] A. Roblin, I. Dubois, F. Grisch et al. Comparison Between Computations and Measurements of a H2/LOX Rocket Motor Plume[C]. 8th AIAA/ASME Joint Thermophysics and heat transfer Conference. AIAA-2002-307:1~10.
[4] Filip Neele, Ric Schleijpen. Electro-Optical Missile Plume Detection[C]. Targets and backgrounds IX: Characterization and Representation. Proceedings of SPIE. Vol 5075, 2003: 270~280.
[5] 宣益民, 韩玉阁. 地面目标与背景的红外特征[M]. 北京: 国防工业出版社, 2004.
[6] 宋自林, 杜益, 黄松. 从“沙漠之狐”行动与科索沃战争看传感器技术与数据融合[J]. 军事电子信息系统专题技术文选. 电子部 28 所. 2000. l3 (1).
[7] 艾雨兵, 郑金华. 舰船电子装备特点及其发展[J]. 情报指挥控制系统与仿真技术, 2004, 16(1): 38~42.
[8] Cuthbertson G M, Seedhouse A. The evolution of a digital dual waveband FLIR[C]. SPIE. 1994, 2269: 665~677.
[9] Clark G A, et al. Land mine detection using multispectral image fusion[C]. Proceedings of the Symposium on Autonomous Vehicles in Mine Countermeasures, Monterey, CA(United States), 1995, 3~7.
[10] 范之国, 高俊, 胡良梅等. 夜视图像自适应彩色化融合算法[J]. 仪器仪表学报. 2006, 27(6): 733~737.
[11] 倪国强, 戴文, 李勇量, 蒲恬. 基于响尾蛇双模式细胞机理的可见光/红外图像彩色融合技术的优势和前景展望[J]. 北京理工大学学报. 2004, 24(2):95~100.
[12] 美国国防部关键技术计划 (1992 财年). 北京: 国防科技信息中心, 1993.
[13] 王宏, 敬忠良, 李建勋. 多分辨率图像融合的研究与发展[J]. 控制理论与应用. 2004, 21(1): 145~151.
[14] 高稚允, 金伟其等. 一种可实时化的多光谱图像融合系统[J]. 光学技术. 1995,4:13~16.
[15] 王毅, 倪国强, 李勇量. 基于 TMS320C6x 的双波段图像高速融合系统[J]. 2002. 7(A):10: 1038~1042.
[16] 倪国强. 多波段图像融合算法研究及其新发展. 光电子技术与信息[J]. 2001,14(5): 11~17.
[17] Peter J B, Edward H A. Merging image through pattern decomposition[C]. SPIE.1985, 575: 173~181.
[18] Toet A, Lodewik J, Ruyven V, et al.. Merging thermal and visual image by a contrast pyramid[J]. Optical Engineering. 1992, 28: 789~792.
[19] Steinnocher K. Application of adaptive filters for multisensoral image fusion[J]. Geoscience and Remote Sensing, 1997, 2: 910~912.
[20] Tery A.Wilson, Steven K.Rogers, Lemuel R.Myers. Perceptual-based hyperspectral image fusion using multiresolution analysis[J]. Optical Engineering. 1995, 34: 3154~3164.
[21] Kundur D, Hatzinakos D. A novel approach to mulitispectral blind image fusion[C]. SPIE. 1997, 3067: 83~93.
[22] Smart S, Scarff L A. Combining visual and IR images for sensor fusion —two approaches[C]. SPIE. 1992, 1668: 102~112.
[23] Yocky D A. Artifacts in Wavelet Image Merging[C].SPIE. 1996, 35(7): 2094~2101.
[24] Ajjimarange P. Neural network model for fusion fo visible and infrared sensor outputs[C]. SPIE. 1988, 1003: 153~160.
[25] Kinser J M. Pulse-coupled image fusion[J]. Optical Engineering. 1997, 36(3): 737~742.
[26] Duane G. Pixel-level sensor fusion for improved object recognition[C]. SPIE. 1998, 931: 180~185.
[27] Streit R L, Luginbuhl T E. Maximum likelihood training of probabilistic neural networks[J]. IEEE Trans. Neur. Networks, 1994 , 5: 764~783.
[28]Peter J B, Edward H A. Merging image through pattern decomposition[C]. SPIE. 1985, 575: 173~181.
[29] 张建奇. 红外物理[M].西安: 西安电子科技大学出版社, 2004: 105~114.
[30] 张敬贤, 李玉丹, 金伟其. 微光与红外成像技术[M]. 北京: 北京理工大学出版社, 1995.
[31] 张智丰, 李向新, 彭群生. 基于大气传输模型的动态目标红外成像仿真[J]. 系统仿真学报.2000,12(9):524-527
[32]王毅, 范伟, 饶瑞中. 典型大气条件下的目标-背景对比度的计算分析[J]. 激光与红外. 2004, 34(5):375~378.
[33]范伟, 王毅, 饶瑞中. 根据大气辐射特征进行目标探测的波段选择[J]. 红外与激光工程. 2002, 34(2):177~182.
[34] 邢强林, 黄惠明, 熊仁生,于 涛. 红外成像探测系统作用距离分析方法研究.光子学报[J].2004,33(4): 894-896.
[35] 王江安, 肖伟岸, 申林. 海空背景下目标红外辐射特征分析.海军工程大学学报[J].2001,13(4): 29-31
[36]吴文健, 胡碧茹, 满亚辉译. 地面目标和背景的热红外特性[M].国防工业出版社.2004
[37]吴北婴, 李卫, 陈洪滨等. 大气传输实用算法[M]. 北京: 气象出版社, 1998.
[38] 吴晗平. 红外辐射大气透过率的工程理论计算方法研究。光学精密工程[J]. 1998,8(4):38-43
[39] 路 远, 凌永顺. 红外辐射大气透射比的简易计算.红外技术[J].2003,25(5): 45-49
[40] 王春勇, 金伟其. 光电成像系统分析中大气传输特性计算的几个问题. 北京理工大学学报[J].2003,23(5): 617-620
[41] Pieter A. Jacobs. 吴文健等译.地面目标和背景的热红外特性[M]. 北京: 国防工业出版社. 2004.
[42] 王江安, 肖伟岸. 基于双波段的目标红外辐射特征分析[J]. 激光与红外. 2001, 31(6):351~353.
[43]金剑秋, 王章野, 彭群生. 相关准则系数下的多光谱图像最优融合方法.计算机辅助设计与图形学学报[J]. 1~6
[44] A.Goldberg, T.Fischer, S.Kennerly. Dual-Band Imaging of Military Targets Using a QWIP Focal Plane Array. EO/IR Technology Branch, Army Research Laboratory, Adelphi, MD 20783
[45] Michelle Tomkinson, Brian Teaney, Jeffrey Olson. Dual Band Sensor Fusion for Urban Target Acquisition. U.S. Army RDECOM CERDEC Night Vision and ElectronicSensors Directorate10221 Burbeck RdFt. Belvoir, Virginia 22060-5806
[46] 汪 强, 刘钢钦. 一种新的基于局部小波统计特性的多聚焦图像融合算法.长沙大学学报[J].2006,20(6): 72-74
[47] 张焕龙, 杨国胜, 王婷,申晓华. 基于局部熵和渐变因子的自适应图像拼接.计算机工程与应用[J].2006,22(3): 72-74
[48] 王江安, 朱向前, 宗思光,肖伟岸. 红外目标特征分析及融合特征提取.传感技术学报[J].2005,6(2): 289-291
[49] 林国余, 张为公. 基于进化规划的最大类间方差的图像分割算法.传感技术学报[J].2006,2(1): 179-182
[50] 李寒松, 鲁传运. 红外舰船小目标的检测. 红外与激光工程[J].2006,35(4): 495-498