基于特征的SAR影像匹配技术研究
|
摘要
图像匹配是遥感影像融合、导航等应用的关键技术之一。随着遥感技术在国民经济、军事中的广泛应用,图像匹配技术也显得越来越重要。光学影像的获取易受光线、天气、昼夜等因素的影响,而合成孔径雷达(Synthetic Aperture Radar,SAR)作为一种主动成像系统具有分辨率高、全天时、全天候等优点。本文针对SAR影像信噪比低、变形大等特点,从影像特征角度,就SAR影像同源匹配、SAR影像与光学影像匹配分别进行了研究。
本文研究的内容和创新点有;
1概要介绍了合成孔径雷达成像的原理和特点,并针对SAR影像的特点研究了SAR影像匹配的步骤和总体方案。
2详细阐述了SAR影像的预处理,探讨了机载SAR影像的几何纠正模型、几何纠正过程。介绍了雷达影像滤波的研究现状,对基于小波变换阈值滤波的SAR影像噪声抑制进行了研究,取得了良好的效果。
3研究了SAR影像同源匹配,引入了一种基于尺度不变特征变换(Scale-InvariantFeature Transform,SIFT)的方法,提取SAR影像的关键点。并针对SAR影像不同于一般的高清晰影像的特点,将SIFT特征匹配与局部灰度匹配相结合,提出了一种去除错配点的相关系数控制方法,对提取的关键点做进一步的控制。实验中匹配概率得到提高,解决了旋转、缩放、扭曲等区别的SAR同源影像高精度匹配问题。
4对于SAR影像与光学影像的匹配,结合小波多分辨率建立影像金字塔,提出了一种改进的Canny算子的边缘提取的算法。在边缘提取的搜索上采用广度优先搜索法,解决了SAR影像噪声过多影响边缘提取的问题。并利用提取出的良好边缘,使用空间距离法结合改进的Hausdorff距离,对SAR影像与光学影像的非同源匹配进行研究,得到了理想的结果。
Image matching is one of the key techniques for such application as the fusion of remote sensing images, technology in automatic aircraft navigation and so on. Optics image acquisition can be easily influenced by lighting, weather and the alternation of day and night, while Synthetic Aperture Radar(SAR) can work all-weather and day-or-night. without being influenced. Therefore, SAR image matching turns out to be the direction, at the same time, the difficulty of matching technique. In view of such characteristics as low S/N ratio and great image distortion of SAR, this thesis makes respective study on same-source SAR image matching and the matching of SAR image with optics image.
Contents and innovations:
1. This article makes a brief introduction of the principle and characteristics of SAR image, and according to these characteristics, studies the stages and general methods of SAR image matching.
2. This article elaborates the preprocessing of SAR image. That is, decreasing the speckle noise of SAR image by wavelet transform with a thresholding, which receives sound result and is applied to SAR image edge detection.
3. To study same-source SAR image matching, the author introduces a method based on Scale Invariant Feature Transform (SIFT) to detect the key-points of SAR image. And as SAR image is different from the ordinary remote sensing image, which features its clearness, the author combines the SIFT feature matching and local gray-scale matching together and puts forward a new controlling method with correlation coefficient for the further elimination of mismatch points. It proves to be very effective to further control the key points with this new method, which resolves the matching problem of same-source SAR images that are distinguished by rotation, scaling and distortion with high precision.
4. With regard to the matching of SAR image with optics image, the author establishes an image pyramid with the help of wavelet multi-resolution transform and brings forward a new edge detection method based on improved Canny algorithm. As for the search of edge detection, the author takes breadth-first search so as to overcome the numerous SAR image speckles. Making good use of the detected edge, the author studies the multi-source matching of SAR image with optics image by integrating weighted Hausdorff distance with space distance, which obtains satisfactory results.
本文研究的内容和创新点有;
1概要介绍了合成孔径雷达成像的原理和特点,并针对SAR影像的特点研究了SAR影像匹配的步骤和总体方案。
2详细阐述了SAR影像的预处理,探讨了机载SAR影像的几何纠正模型、几何纠正过程。介绍了雷达影像滤波的研究现状,对基于小波变换阈值滤波的SAR影像噪声抑制进行了研究,取得了良好的效果。
3研究了SAR影像同源匹配,引入了一种基于尺度不变特征变换(Scale-InvariantFeature Transform,SIFT)的方法,提取SAR影像的关键点。并针对SAR影像不同于一般的高清晰影像的特点,将SIFT特征匹配与局部灰度匹配相结合,提出了一种去除错配点的相关系数控制方法,对提取的关键点做进一步的控制。实验中匹配概率得到提高,解决了旋转、缩放、扭曲等区别的SAR同源影像高精度匹配问题。
4对于SAR影像与光学影像的匹配,结合小波多分辨率建立影像金字塔,提出了一种改进的Canny算子的边缘提取的算法。在边缘提取的搜索上采用广度优先搜索法,解决了SAR影像噪声过多影响边缘提取的问题。并利用提取出的良好边缘,使用空间距离法结合改进的Hausdorff距离,对SAR影像与光学影像的非同源匹配进行研究,得到了理想的结果。
Image matching is one of the key techniques for such application as the fusion of remote sensing images, technology in automatic aircraft navigation and so on. Optics image acquisition can be easily influenced by lighting, weather and the alternation of day and night, while Synthetic Aperture Radar(SAR) can work all-weather and day-or-night. without being influenced. Therefore, SAR image matching turns out to be the direction, at the same time, the difficulty of matching technique. In view of such characteristics as low S/N ratio and great image distortion of SAR, this thesis makes respective study on same-source SAR image matching and the matching of SAR image with optics image.
Contents and innovations:
1. This article makes a brief introduction of the principle and characteristics of SAR image, and according to these characteristics, studies the stages and general methods of SAR image matching.
2. This article elaborates the preprocessing of SAR image. That is, decreasing the speckle noise of SAR image by wavelet transform with a thresholding, which receives sound result and is applied to SAR image edge detection.
3. To study same-source SAR image matching, the author introduces a method based on Scale Invariant Feature Transform (SIFT) to detect the key-points of SAR image. And as SAR image is different from the ordinary remote sensing image, which features its clearness, the author combines the SIFT feature matching and local gray-scale matching together and puts forward a new controlling method with correlation coefficient for the further elimination of mismatch points. It proves to be very effective to further control the key points with this new method, which resolves the matching problem of same-source SAR images that are distinguished by rotation, scaling and distortion with high precision.
4. With regard to the matching of SAR image with optics image, the author establishes an image pyramid with the help of wavelet multi-resolution transform and brings forward a new edge detection method based on improved Canny algorithm. As for the search of edge detection, the author takes breadth-first search so as to overcome the numerous SAR image speckles. Making good use of the detected edge, the author studies the multi-source matching of SAR image with optics image by integrating weighted Hausdorff distance with space distance, which obtains satisfactory results.
引文
[1]于秋则.基于边缘特征的SAR影像与光学影像的匹配[J].雷达科学与技术,2003,12.
[2]Mehrdad S.Synthetic aperture radar signal processing with MATLAB Algorithms[M].NewYork;A Wiley-Interscience Publication,1999.
[3]刘永坦.雷达成像技术[M].哈尔滨;哈尔滨工业大学出版社,1999.
[4]S.D.Madsen.Estimating the Doppler centroid of SAR data.IEEE Trans.On AES,1989,25(2);134-140.
[5]曲长文.机载SAR发展概况[J].现代雷达.2002,01.
[6]Giorgio Franceschetti,Riccando Lanari.Synthetic Aperture Radar Processing[M].CRC Press.1999.
[7]马元申.星载合成子L径雷达的发展及其作战应用[J].现代雷达,2002,01.
[8]曲长文.机载SAR发展概况[J].遥感技术与应用,2001,12.
[9]ELACH C,et al.Spacebome synthetic aperture imaging radars;applications,techniques and echnology[J].Proceedings of IEEE,1982,70(10);1174-1209.
[10]袁孝康.星载合成孔径雷达的原理、组成和性能[J].上海航天,1997,14(1);49-55.
[11]Huang J,et al.A inflatable L-band microstrip SAR array[J].1998 IEEE AP-S Int.Symp.2100-2103.
[12]FreemanA.SAR calibration;an overview[J].IEEE Trans.On Geoscience and Remote Sensing,1992,30(6);1107-1121.
[13]John C.Curlander and Robert N.Mcdonough Synthetic Aperture Radar Systems and Signal Peocessing[M].A Wiley-Interscience Publication(USA),1991,8(6);1-66.
[14]于秋则.合成孔径雷达图像匹配导航技术研究[D].华中科技大学.2004,05.
[15]http;//www.sandia.gov/RADA1Vsarapps.html.
[16]廖明生,林珲.雷达干涉测量一原理与信号处理基础[M],北京;测绘出版社,2003.
[17]Dare P,Dowman Ⅰ.An improved model for automatic feature-based registration of SAR and SPOT images[J].ISPRS(International Society for Photogrammetry and Remote Sensing)Journal of Photogrammetry&Remote Sensing.2001,(56);13-28.
[18]Huang Yong,et al.An image matching algorithm and implement for SAR images[J].Signal Processing.2003,19(2);179-183.
[19]马昕.多传感器惯性组合导航技术研究[D].南京;南京航空航天大学博士论文,1998.
[20]于秋则.合成孔径雷达图像匹配导航技术研究[D].华中科技大学.2004,05.
[21]朱述龙,张占睦.遥感图象获取与分析[M].北京;科学出版社.2000,04.
[22]黄勇,王建国,黄顺吉.图像匹配算法及实现[J].信号处理.2003,19(2).
[23]康欣,韩崇昭,杨艺.基于结构的SAR影像配准[J].系统仿真学报.2006,18(5).
[24]A Goshtasby,G Stockman,C Page.A region-based approach to digital image registration with subpixel accuracy[J].IEEETrans.Geosci.Remote Sensing(S0196-2892),1986,24(3);390-399.
[25]靳国旺.航天INSAR提取DEM的研究[D].郑州;解放军信息工程大学硕士论文,2003.
[26]韩春明.SAR影像斑点滤波研究[D].北京;中国科学院博士论文,2003.
[27]蒋祥玉.基于灰度共生矩阵的景象匹配算法研究[D].哈尔滨;哈尔滨工业大学硕士论文,2004.07.
[28]邓炜.SAR影像处理方法的研究[D].西安;西北工业大学博士学位论文,2000.
[29]David Lowe.Distinctive image features from scale-invariant keypoints[M].2001.02[EB/OL].
[30]David Lowe.Object recognition from local scale-invariant features[EB/OL].
[31]David Lowe.Local feature view clustering for 3D object recognition[EB/OL].
[32]赵辉.SIFT算法讲义[R].
[33]http;//www.cs.ubc.ca/-lowe/.
[34]陈鹰.基于边缘特征的SAR图像与光学影像的匹配[J].雷达科学与技术,2003,12.
[35]Xiaolong Dai,Siamak Khorram.A Feature-Based Image Registration Algorithm Using Improved Chain-Code Representation Combined With Invariant Moments.IEEE Trans.On GRS,Vol37,No.5,Sep 1999.
[36]H Li,B S Manjunath,and S K.Mitra.A Contour Based Approach to Multi-Sensor Image Registration.IEEE Trans.GRS,Vol.4,Pp.320-334,Marl995.
[37]杨卫东,左峥嵘,张天序.基于检测识别的实孔径雷达景象匹配定位方法.华中科技大学学报(自然 科学版),Vol.33,No.2,Feb.2005.
[38]巫兆聪,方圣辉.基于分形理论的SAR影像边缘检测[J].武汉测绘科技大学学报,2000,25(4);334-337.
[39]Burns.Extracting Straight Lines.IEEE PAMI 1986,8(4);425-455.
[40]Willian T.Freeman and Edward H.Adelson The Design and Use of Stcerable Filtres.IEEE Trans PAMI,1991(13);891-906.
[41]华继钊.Canny算法的改进[J].微机发展.2005.15(4);24-26.
[42]李杰,丁宣浩.一种改进的小波阈值消噪方法[J].广西工学院学报.2006.17(2).56-59.
[43]陈岚岚等.Hausdorff距离在图像匹配中的应用[J].现代电子技术,2002.9.
[44]Huttenlocher D P,Klanderm,G A,Rucklidge W J.Comparing images using the Hausdorffdistance.IEEE Trans.Pattern Anal.Machine Intell,1993,15(3);850-863.
[45]Suchandt.Suchandt-An Airborne SAR Experiment for Ground Moving Target Identification.
[46]Luis M.T.Carvalho,Leila M.G.Fonseca,Fionn Murtagh,Jan G.P.W.Clevers,Changes at multiple spatial scales[C],ISPRS 2000.
[47]Franz Kurz and Olaf Hellwich,Empirical Estimation of Vegetation Parameters Sing Multisensor Data Fusion[C],ISPRS 2000.
[48]Johannes Leebmann,Dietmar Kunz,Fusion of Satellite Imagery and Digital Topographic Database Based on Disjoint Objects[C],ISPRS 2000.
[49]U.S.Department of Defense,Data fusion subpanel of the joint directors of laboratories[R],Technical Panel for C3,Data fusion lexicon,1991.
[50]Hall,D.L.1992.Mathematical techniques in multisensor data fusion[R].Boston;Artech House.
[51]Jong-Hyun Park,Ryutaro Tateishi,Ketut Wikantika.Multisensor Data Fusion Using Multiresoludion Analysis(MRA)[C],ISPRS 2000.
[52]A.D.Ventura,A.Rampini.And R.SeheUini,Image Registration By Recognition of Corresponding Struttures,IEEE Trans.Grs,Vol.28,Pp.305-314,May 1990.
[53]Stockman G.Matching images to models for registration and object detetion via clustering[J].IEEE Trans.On PAMI,1982,4(3);229-234.
[54]刘贵喜,赵曙光,陈文锦.双正交小波变换多分辨率图像融合方法.2004,4,第31卷.
[55]Frost V S,Stiles J A,Shanmugan K S,et al.A Modal for Radar Images and Its Application to Adaptive Digital Filtering of Multiplieative Noise[J].IEEE Transactinns on Pattern Analysis and Machine Intelligence,1982,PAMI24(2);157.166.
[56]Touzi R,Lopes A,Bousquet P.A Statistical and Geometrical Edge Detector for SAR Images[J],IEEE Transactions of Geoscience and Remote Sensing,1988,26(6);764,773.
[57]魏钟铨等.合成孔径雷达卫星[M].北京;科学出版社,2001.
[58]李郁,闰敬文.一种基于小波分析的SAR影像增强去噪方法.计算机工程与设计.2005,2,第25卷.
[59]Shen Gui-ming.An new image denoising method using wavelet transformation technique[J].Journal of Xiamen University,2000,39(5);612-616.
[60]D P Huttenlocher,G A Klanderman,W J Rucklidge.Comparing images using the Hausdorf distance[J].IEEE Tram Pattern Anal.Machine lntell,1993,15(9);850-863.
[61]Jane You.A wavelet-based coarse-to-fine image matching scheme in a parallet virtual machine environment[J].IEEE Transactions On image processing,2000,9(9);1547-1559.
[62]O K Kwon,D G Sim,R H Park.Robust Hausdorfdistance matching algorithms using pyramidal structures[J].Pattern Recognition.2001,34(7);2005-2013.
[2]Mehrdad S.Synthetic aperture radar signal processing with MATLAB Algorithms[M].NewYork;A Wiley-Interscience Publication,1999.
[3]刘永坦.雷达成像技术[M].哈尔滨;哈尔滨工业大学出版社,1999.
[4]S.D.Madsen.Estimating the Doppler centroid of SAR data.IEEE Trans.On AES,1989,25(2);134-140.
[5]曲长文.机载SAR发展概况[J].现代雷达.2002,01.
[6]Giorgio Franceschetti,Riccando Lanari.Synthetic Aperture Radar Processing[M].CRC Press.1999.
[7]马元申.星载合成子L径雷达的发展及其作战应用[J].现代雷达,2002,01.
[8]曲长文.机载SAR发展概况[J].遥感技术与应用,2001,12.
[9]ELACH C,et al.Spacebome synthetic aperture imaging radars;applications,techniques and echnology[J].Proceedings of IEEE,1982,70(10);1174-1209.
[10]袁孝康.星载合成孔径雷达的原理、组成和性能[J].上海航天,1997,14(1);49-55.
[11]Huang J,et al.A inflatable L-band microstrip SAR array[J].1998 IEEE AP-S Int.Symp.2100-2103.
[12]FreemanA.SAR calibration;an overview[J].IEEE Trans.On Geoscience and Remote Sensing,1992,30(6);1107-1121.
[13]John C.Curlander and Robert N.Mcdonough Synthetic Aperture Radar Systems and Signal Peocessing[M].A Wiley-Interscience Publication(USA),1991,8(6);1-66.
[14]于秋则.合成孔径雷达图像匹配导航技术研究[D].华中科技大学.2004,05.
[15]http;//www.sandia.gov/RADA1Vsarapps.html.
[16]廖明生,林珲.雷达干涉测量一原理与信号处理基础[M],北京;测绘出版社,2003.
[17]Dare P,Dowman Ⅰ.An improved model for automatic feature-based registration of SAR and SPOT images[J].ISPRS(International Society for Photogrammetry and Remote Sensing)Journal of Photogrammetry&Remote Sensing.2001,(56);13-28.
[18]Huang Yong,et al.An image matching algorithm and implement for SAR images[J].Signal Processing.2003,19(2);179-183.
[19]马昕.多传感器惯性组合导航技术研究[D].南京;南京航空航天大学博士论文,1998.
[20]于秋则.合成孔径雷达图像匹配导航技术研究[D].华中科技大学.2004,05.
[21]朱述龙,张占睦.遥感图象获取与分析[M].北京;科学出版社.2000,04.
[22]黄勇,王建国,黄顺吉.图像匹配算法及实现[J].信号处理.2003,19(2).
[23]康欣,韩崇昭,杨艺.基于结构的SAR影像配准[J].系统仿真学报.2006,18(5).
[24]A Goshtasby,G Stockman,C Page.A region-based approach to digital image registration with subpixel accuracy[J].IEEETrans.Geosci.Remote Sensing(S0196-2892),1986,24(3);390-399.
[25]靳国旺.航天INSAR提取DEM的研究[D].郑州;解放军信息工程大学硕士论文,2003.
[26]韩春明.SAR影像斑点滤波研究[D].北京;中国科学院博士论文,2003.
[27]蒋祥玉.基于灰度共生矩阵的景象匹配算法研究[D].哈尔滨;哈尔滨工业大学硕士论文,2004.07.
[28]邓炜.SAR影像处理方法的研究[D].西安;西北工业大学博士学位论文,2000.
[29]David Lowe.Distinctive image features from scale-invariant keypoints[M].2001.02[EB/OL].
[30]David Lowe.Object recognition from local scale-invariant features[EB/OL].
[31]David Lowe.Local feature view clustering for 3D object recognition[EB/OL].
[32]赵辉.SIFT算法讲义[R].
[33]http;//www.cs.ubc.ca/-lowe/.
[34]陈鹰.基于边缘特征的SAR图像与光学影像的匹配[J].雷达科学与技术,2003,12.
[35]Xiaolong Dai,Siamak Khorram.A Feature-Based Image Registration Algorithm Using Improved Chain-Code Representation Combined With Invariant Moments.IEEE Trans.On GRS,Vol37,No.5,Sep 1999.
[36]H Li,B S Manjunath,and S K.Mitra.A Contour Based Approach to Multi-Sensor Image Registration.IEEE Trans.GRS,Vol.4,Pp.320-334,Marl995.
[37]杨卫东,左峥嵘,张天序.基于检测识别的实孔径雷达景象匹配定位方法.华中科技大学学报(自然 科学版),Vol.33,No.2,Feb.2005.
[38]巫兆聪,方圣辉.基于分形理论的SAR影像边缘检测[J].武汉测绘科技大学学报,2000,25(4);334-337.
[39]Burns.Extracting Straight Lines.IEEE PAMI 1986,8(4);425-455.
[40]Willian T.Freeman and Edward H.Adelson The Design and Use of Stcerable Filtres.IEEE Trans PAMI,1991(13);891-906.
[41]华继钊.Canny算法的改进[J].微机发展.2005.15(4);24-26.
[42]李杰,丁宣浩.一种改进的小波阈值消噪方法[J].广西工学院学报.2006.17(2).56-59.
[43]陈岚岚等.Hausdorff距离在图像匹配中的应用[J].现代电子技术,2002.9.
[44]Huttenlocher D P,Klanderm,G A,Rucklidge W J.Comparing images using the Hausdorffdistance.IEEE Trans.Pattern Anal.Machine Intell,1993,15(3);850-863.
[45]Suchandt.Suchandt-An Airborne SAR Experiment for Ground Moving Target Identification.
[46]Luis M.T.Carvalho,Leila M.G.Fonseca,Fionn Murtagh,Jan G.P.W.Clevers,Changes at multiple spatial scales[C],ISPRS 2000.
[47]Franz Kurz and Olaf Hellwich,Empirical Estimation of Vegetation Parameters Sing Multisensor Data Fusion[C],ISPRS 2000.
[48]Johannes Leebmann,Dietmar Kunz,Fusion of Satellite Imagery and Digital Topographic Database Based on Disjoint Objects[C],ISPRS 2000.
[49]U.S.Department of Defense,Data fusion subpanel of the joint directors of laboratories[R],Technical Panel for C3,Data fusion lexicon,1991.
[50]Hall,D.L.1992.Mathematical techniques in multisensor data fusion[R].Boston;Artech House.
[51]Jong-Hyun Park,Ryutaro Tateishi,Ketut Wikantika.Multisensor Data Fusion Using Multiresoludion Analysis(MRA)[C],ISPRS 2000.
[52]A.D.Ventura,A.Rampini.And R.SeheUini,Image Registration By Recognition of Corresponding Struttures,IEEE Trans.Grs,Vol.28,Pp.305-314,May 1990.
[53]Stockman G.Matching images to models for registration and object detetion via clustering[J].IEEE Trans.On PAMI,1982,4(3);229-234.
[54]刘贵喜,赵曙光,陈文锦.双正交小波变换多分辨率图像融合方法.2004,4,第31卷.
[55]Frost V S,Stiles J A,Shanmugan K S,et al.A Modal for Radar Images and Its Application to Adaptive Digital Filtering of Multiplieative Noise[J].IEEE Transactinns on Pattern Analysis and Machine Intelligence,1982,PAMI24(2);157.166.
[56]Touzi R,Lopes A,Bousquet P.A Statistical and Geometrical Edge Detector for SAR Images[J],IEEE Transactions of Geoscience and Remote Sensing,1988,26(6);764,773.
[57]魏钟铨等.合成孔径雷达卫星[M].北京;科学出版社,2001.
[58]李郁,闰敬文.一种基于小波分析的SAR影像增强去噪方法.计算机工程与设计.2005,2,第25卷.
[59]Shen Gui-ming.An new image denoising method using wavelet transformation technique[J].Journal of Xiamen University,2000,39(5);612-616.
[60]D P Huttenlocher,G A Klanderman,W J Rucklidge.Comparing images using the Hausdorf distance[J].IEEE Tram Pattern Anal.Machine lntell,1993,15(9);850-863.
[61]Jane You.A wavelet-based coarse-to-fine image matching scheme in a parallet virtual machine environment[J].IEEE Transactions On image processing,2000,9(9);1547-1559.
[62]O K Kwon,D G Sim,R H Park.Robust Hausdorfdistance matching algorithms using pyramidal structures[J].Pattern Recognition.2001,34(7);2005-2013.