基于Haar特征的高分辨率遥感影像地物识别方法研究
|
摘要
地物信息的自动提取一直是遥感影像处理、土地利用分类及动态监测等领域的研究热点，目前已经形成了一些比较成熟的地物提取方法，如监督和非监督分类、面向对象信息提取等，均可得到较好的效果；其中，BP神经网络和支持向量机则非常适合于高分辨率遥感影像上的小地物识别，但仍然存在算法效率低，分类器训练速度较慢、依赖经验性等问题。本文以影像上的汽车为例，在利用WorldView-2全色影像数据进行地物识别研究的基础上，提出了在Microsoft Visual C++平台下，借助OpenCV函数库，从影像上提取汽车样本图像的Haar特征，并采用Adaboost算法训练分类器，在高分辨率遥感影像上进行小地物的自动识别、提取的新方法，通过具体的实验效果探讨了这种方法的适用性与可行性，总结归纳出了完整的技术方法流程及相关程序的说明；在此基础之上，深入研究了影像分辨率改变对汽车识别效果的影响，通过实验论述了遥感影像应用方面的尺度问题。
本文的主要研究内容和结论包括以下三个方面：
（1）在总结前人研究的基础上，利用WorldView-2全色影像，以汽车为例，通过实验论证了提取目标地物的Haar特征，采用Adaboost算法训练分类器的目标识别方法可以应用于高分辨率遥感影像上的小地物识别，拓展了算法的应用领域；
（2）归纳总结了完整的实验技术方法流程，并对相关识别程序代码进行了说明，为遥感影像上特定物体的识别提供参考；
（3）研究了影像分辨率改变对汽车识别效果的影响，在一定程度上论证了遥感影像应用的尺度问题，指出要识别一定尺度的地物所需影像的分辨率必须满足一定的条件。
Automatic feature extraction has always been a hot topic in the area ofremote sense image processing, land us classification and land use dynamicmonitoring. There are many well developed methods, such as classification withand without monitoring, object-oriented feature extraction, among which BPNeural Networks and Support Vector Machine methods are better for smallobject detection in the high resolution remote sense image. In general, theabove methods can produce good results, but most of them have the problems oflow efficiency, low classifier-training speed and experience dependent. Basedon the study of vehicle targets detection in WorldView-2full-color image, weproposed a new approach in platform of Microsoft Visual C++for automaticallyidentifying small targets in high-resolution remote sense image. This methoduses OpenCV Library to extract Haar features of samples and then trains theclassifier with Adaboost Algorithm. We discussed the applicability andfeasibility of this method through experimental results and presented thecomplete technical procedure and program scheme. Also we studied the effectof image resolution on the detection of vehicle targets and discussed thepossible scale problem in application.
The main content of this article can be summarized as follow:
(1) Based on the previous studies, a new method of object identification usingHaar features and Adaboost classification training algorithm is proposed, whichcan be used for small object detection in the high resolution remote sense image.Expand the applications of Adaboost Algorithm.
(2) The complete technical procedure and program scheme are summarized forfurther application of special objects detection in high-resolution remote senseimage.
(3) The effect of image resolution on the detection of vehicle targets is studied,as well as the necessary resolution for the detected targets of given scale.
本文的主要研究内容和结论包括以下三个方面：
（1）在总结前人研究的基础上，利用WorldView-2全色影像，以汽车为例，通过实验论证了提取目标地物的Haar特征，采用Adaboost算法训练分类器的目标识别方法可以应用于高分辨率遥感影像上的小地物识别，拓展了算法的应用领域；
（2）归纳总结了完整的实验技术方法流程，并对相关识别程序代码进行了说明，为遥感影像上特定物体的识别提供参考；
（3）研究了影像分辨率改变对汽车识别效果的影响，在一定程度上论证了遥感影像应用的尺度问题，指出要识别一定尺度的地物所需影像的分辨率必须满足一定的条件。
Automatic feature extraction has always been a hot topic in the area ofremote sense image processing, land us classification and land use dynamicmonitoring. There are many well developed methods, such as classification withand without monitoring, object-oriented feature extraction, among which BPNeural Networks and Support Vector Machine methods are better for smallobject detection in the high resolution remote sense image. In general, theabove methods can produce good results, but most of them have the problems oflow efficiency, low classifier-training speed and experience dependent. Basedon the study of vehicle targets detection in WorldView-2full-color image, weproposed a new approach in platform of Microsoft Visual C++for automaticallyidentifying small targets in high-resolution remote sense image. This methoduses OpenCV Library to extract Haar features of samples and then trains theclassifier with Adaboost Algorithm. We discussed the applicability andfeasibility of this method through experimental results and presented thecomplete technical procedure and program scheme. Also we studied the effectof image resolution on the detection of vehicle targets and discussed thepossible scale problem in application.
The main content of this article can be summarized as follow:
(1) Based on the previous studies, a new method of object identification usingHaar features and Adaboost classification training algorithm is proposed, whichcan be used for small object detection in the high resolution remote sense image.Expand the applications of Adaboost Algorithm.
(2) The complete technical procedure and program scheme are summarized forfurther application of special objects detection in high-resolution remote senseimage.
(3) The effect of image resolution on the detection of vehicle targets is studied,as well as the necessary resolution for the detected targets of given scale.
引文
[1]周成虎,骆剑承等.高分辨率卫星遥感影像地学计算.科学出版社,2009
[2]阎守邕,刘亚岚等.遥感影像群判读理论与方法.海洋出版社,2007
[3]明冬萍,骆剑承等.高分辨率遥感影像信息提取与目标识别技术研究.测绘科学,2005,30(3):18-20
[4]熊秩群,吴健平.面向对象的城市绿地信息提取方法研究.华东师范大学学报（自然科学版）,2006,4:84-90
[5]李彦,沈旭坤.基于高斯模型的遥感影像目标识别方法的初探.系统仿真学报,2009(S1):57-60
[6]王志伟,关泽群.基于多不变量特征的遥感影像小地物识别.遥感信息,2005(6):17-19
[7]Jens Leitloff, Stefan Hinz. Vehicle Detection in Very High Resolution Satellite Images of CityArea. Geosience and Remote Sensing,2010,7(48):2795-2806
[8]邓书斌. ENVI遥感图像处理方法.科学出版社,2011
[9]黄慧萍,吴炳方等.高分辨率也能共享城市绿地快速提取技术与应用.遥感学报,2004,1(8):68-74
[10]黎展荣,王龙波.利用高分辨率影像计算城市绿地覆盖率.测绘通报,2006(12):51-53
[11]张春晓,侯伟,刘翔,等.基于面向对象和影像认知的遥感影像分类方法——以都江堰向峨乡区域为例.测绘通报,2010(8):11-14
[12]Gray Bradski, Adrian Kaebler. Learning OpenCV. O’Reilly Media,Inc,2008
[13]许宜申,顾济华等.基于改进BP神经网络的手写字符识别.通信技术,2011,5(44):106-109
[14]王俊芳,杨武年,李玉霞等.多种特征信息在高分辨率遥感影像目标识别中的应用研究.四川测绘,2006,29(4):156-158
[15]Paul Viola, Michael Jones. Robust Real-time Object Detection. Statistical and ComputationalTheories of Vision,2001:1-22
[16]郭磊,王秋光.Adaboost人脸检测算法研究及OpenCV实现.哈尔滨理工大学学报,2009,14(5):123-126
[17]朱文佳,戚飞虎.基于Gentle Adaboost的行人检测.中国图像图形学报,2007,2(10):1905-1908
[18]魏武,张亚楠,武林林.基于遗传算法的改进AdaBoost算法在汽车识别中的应用.公路交通科技,2010,27(2):114-118
[19]刘慧英,王小波.基于OpenCV的车辆轮廓检测.科学技术与工程,2010,10(12):2987-2991
[20]蒋李兵.基于高分辨率光学遥感图像的舰船目标检测方法研究.长沙:国防科技大学,2006:6-9
[21]梅建新.基于支持向量机的高分辨率遥感影像的目标检测研究.武汉:武汉大学,2004:6-16
[22]Y.‐K. Lai,C.‐C. J. Kuo. A Haar Wavelet Approach to Compressed Image QualityMeasurement[J]. Journal of Visual Communication and Image Representation,2000,(11):17‐40
[23] FREUND Y,SCHAPIRE R. A Short Introduction to Boosting. Journal of Japanese Society forArtificial Intelligence,1999,14(5):771-780
[24]Paul Viola, Michael Jones. Rapid Object Detection using a Boosted Cascade of SimpleFeatures. Computer Vision and Pattern Recognition,2001,14(5):1-9
[25]赖敏.基于Adaboost迭代学习的支持向量机分类算法.重庆:重庆师范大学,2010:2-3
[26]Y. Freund, R. E. Schapire.“Experiment with a New Boosting Algorithm”, in MachineLearning: Proceedings of the Thirteenth International Conference(Morgan Kauman, San Francisco,1996),148-156
[27]涂承胜,刁力力,鲁明羽等.Boosting家族Adaboost系列代表算法.计算机科学,2003,30(3):30-34
[28]赵万鹏,古乐野.基于Adaboost的手写体数字识别.计算机应用,2005(10):2413-2414
[29]龚健雅,姚璜,沈心.利用Adaboost算法进行高分辨率影像的面向对象分类.武汉大学学报(信息科学版),2010,35(12):1440-1443
[30]张莹,李勇平,敖新宇.基于OpenCV的通用人脸检测模块设计.计算机工程与科学,2011,33(1):97-101
[31]黎松,平西建,丁益洪.开放源代码的计算机视觉类库OpenCV的应用.计算机应用与软件,2005,22(8):134-136
[32]陈磊.计算机视觉类库OpenCV在VC中的应用.软件时空,2007,23(4):209-210
[33]李雅莉,周文杰.基于OpenCV和VC6.0的数据监控系统设计.电子元器件应用,2011,13(2):44-45
[34] S. Suchandt, H. Runge, H. Breit, A. Kotenkov, D. Weihing, and S. Hinz,“Trafficmeasurement with TerraSAR-X: Processing system overview and first results,” Proc. EUSAR,Friedrichshafen, Germany,2008. http://elib.dlr.de/53757/
[35]凌春丽,朱兰艳,吴俐民.WorldView-2影像林地信息提取的研究与实现.测绘科学,2010,35(5):205-207
[36]阳牧男,姜贞白.WorldViewII卫星影像融合方法的探讨.测绘,2011,34(5):205-207
[37]梅安新,彭望琭.遥感导论.高等教育出版社,2005
[38]L. Wang, W. P. Sousa, P. Gong, G. S. Biging. Comparison of IKONOS and QuickBird imagesfor mapping mangrove species on the Caribbean coast of Panama.Remote Sensing ofEnvironment,2004,91:432-440.
[39]李明阳,申世广,吴翼等.南京紫金山风景林多情境规划方法研究.南京林业大学学报(自然科学版),2007,31(5):29-33
[40]S.E. Franklin, M. A. Wulder, G.R. Gerylo. Texture Analysis of IKONOS Panchromatic Datafor Douglas-fir Forest Age Class Separability in British Columbia.International Journal ofRemote Sensing.2001,22(13):2627—2632.
[41]吴见,彭道黎.基于面向对象的QuickBird影像退耕地树冠信息提取.光谱学与光谱分析,2010,30(9):2533-2536
[42]叶时平,陈超祥,魏玉璋等.基于灰度和纹理特征的QuickBird影像中土地利用信息的提取.通信学报,2008,29(8):129-135
[43]Mena J B, Malpica J A. An automatic method for road extraction in rural and semi-urban arcasstarting from high resolution satellite imagery.Pattern Recognition Letters,2005,26:1201-1220.
[44]翁永玲,范兴旺,胡伍生等.基于QuickBird数据的输电线路径优选中地表信息提取.东南大学学报(自然科学版),2010,40(3):587-592
[45]张海涛,贾光军,虞欣.基于GeoEye-1卫星影像的立体测图技术研究.测绘通报,2010(12):43-46
[46]张治清,何宗.GeoEye-1多光谱与全色影像融合的适应性及质量评价研究.西南师范大学学报(自然科学版),2011,36(1):203-208
[47]虞欣,李和军,贾光军等.GeoEye-1卫星影像定向精度初步分析.测绘通报,2011(1):28-30
[48]黄慧萍,吴炳方.地物大小、对象尺度、影像分辨率的关系分析.遥感技术与应用,2006,21(3):243-248
[49]杜永明,秦其明.不同分辨率对遥感影像中识别人造地物的影响.遥感技术与应用,2001,16(4):214-217
[50]苏理宏,李小文,黄裕霞.遥感尺度问题研究进展.地球科学进展,2001,16(4):544-548
[51]柴渊,李万东.土地利用动态遥感监测技术与方法.地质出版社,2011
[52]李海洋.遥感图像分类方法综述.林业科技情报,2008,40(1):4-5
[53]刘瑞祯,于仕琪.OpenCV教程基础篇.北京航空航天大学出版社,2007
[54]卫亚星,王莉雯.遥感图像增强方法分析.测绘空间地理信息,2006,29(2):3-7
[55]http://www.opencv.org.cn/index.php/VC_2008_Express%E4%B8%8B%E5%AE%89%E8%A3%85OpenCV2.0/2.1
[2]阎守邕,刘亚岚等.遥感影像群判读理论与方法.海洋出版社,2007
[3]明冬萍,骆剑承等.高分辨率遥感影像信息提取与目标识别技术研究.测绘科学,2005,30(3):18-20
[4]熊秩群,吴健平.面向对象的城市绿地信息提取方法研究.华东师范大学学报（自然科学版）,2006,4:84-90
[5]李彦,沈旭坤.基于高斯模型的遥感影像目标识别方法的初探.系统仿真学报,2009(S1):57-60
[6]王志伟,关泽群.基于多不变量特征的遥感影像小地物识别.遥感信息,2005(6):17-19
[7]Jens Leitloff, Stefan Hinz. Vehicle Detection in Very High Resolution Satellite Images of CityArea. Geosience and Remote Sensing,2010,7(48):2795-2806
[8]邓书斌. ENVI遥感图像处理方法.科学出版社,2011
[9]黄慧萍,吴炳方等.高分辨率也能共享城市绿地快速提取技术与应用.遥感学报,2004,1(8):68-74
[10]黎展荣,王龙波.利用高分辨率影像计算城市绿地覆盖率.测绘通报,2006(12):51-53
[11]张春晓,侯伟,刘翔,等.基于面向对象和影像认知的遥感影像分类方法——以都江堰向峨乡区域为例.测绘通报,2010(8):11-14
[12]Gray Bradski, Adrian Kaebler. Learning OpenCV. O’Reilly Media,Inc,2008
[13]许宜申,顾济华等.基于改进BP神经网络的手写字符识别.通信技术,2011,5(44):106-109
[14]王俊芳,杨武年,李玉霞等.多种特征信息在高分辨率遥感影像目标识别中的应用研究.四川测绘,2006,29(4):156-158
[15]Paul Viola, Michael Jones. Robust Real-time Object Detection. Statistical and ComputationalTheories of Vision,2001:1-22
[16]郭磊,王秋光.Adaboost人脸检测算法研究及OpenCV实现.哈尔滨理工大学学报,2009,14(5):123-126
[17]朱文佳,戚飞虎.基于Gentle Adaboost的行人检测.中国图像图形学报,2007,2(10):1905-1908
[18]魏武,张亚楠,武林林.基于遗传算法的改进AdaBoost算法在汽车识别中的应用.公路交通科技,2010,27(2):114-118
[19]刘慧英,王小波.基于OpenCV的车辆轮廓检测.科学技术与工程,2010,10(12):2987-2991
[20]蒋李兵.基于高分辨率光学遥感图像的舰船目标检测方法研究.长沙:国防科技大学,2006:6-9
[21]梅建新.基于支持向量机的高分辨率遥感影像的目标检测研究.武汉:武汉大学,2004:6-16
[22]Y.‐K. Lai,C.‐C. J. Kuo. A Haar Wavelet Approach to Compressed Image QualityMeasurement[J]. Journal of Visual Communication and Image Representation,2000,(11):17‐40
[23] FREUND Y,SCHAPIRE R. A Short Introduction to Boosting. Journal of Japanese Society forArtificial Intelligence,1999,14(5):771-780
[24]Paul Viola, Michael Jones. Rapid Object Detection using a Boosted Cascade of SimpleFeatures. Computer Vision and Pattern Recognition,2001,14(5):1-9
[25]赖敏.基于Adaboost迭代学习的支持向量机分类算法.重庆:重庆师范大学,2010:2-3
[26]Y. Freund, R. E. Schapire.“Experiment with a New Boosting Algorithm”, in MachineLearning: Proceedings of the Thirteenth International Conference(Morgan Kauman, San Francisco,1996),148-156
[27]涂承胜,刁力力,鲁明羽等.Boosting家族Adaboost系列代表算法.计算机科学,2003,30(3):30-34
[28]赵万鹏,古乐野.基于Adaboost的手写体数字识别.计算机应用,2005(10):2413-2414
[29]龚健雅,姚璜,沈心.利用Adaboost算法进行高分辨率影像的面向对象分类.武汉大学学报(信息科学版),2010,35(12):1440-1443
[30]张莹,李勇平,敖新宇.基于OpenCV的通用人脸检测模块设计.计算机工程与科学,2011,33(1):97-101
[31]黎松,平西建,丁益洪.开放源代码的计算机视觉类库OpenCV的应用.计算机应用与软件,2005,22(8):134-136
[32]陈磊.计算机视觉类库OpenCV在VC中的应用.软件时空,2007,23(4):209-210
[33]李雅莉,周文杰.基于OpenCV和VC6.0的数据监控系统设计.电子元器件应用,2011,13(2):44-45
[34] S. Suchandt, H. Runge, H. Breit, A. Kotenkov, D. Weihing, and S. Hinz,“Trafficmeasurement with TerraSAR-X: Processing system overview and first results,” Proc. EUSAR,Friedrichshafen, Germany,2008. http://elib.dlr.de/53757/
[35]凌春丽,朱兰艳,吴俐民.WorldView-2影像林地信息提取的研究与实现.测绘科学,2010,35(5):205-207
[36]阳牧男,姜贞白.WorldViewII卫星影像融合方法的探讨.测绘,2011,34(5):205-207
[37]梅安新,彭望琭.遥感导论.高等教育出版社,2005
[38]L. Wang, W. P. Sousa, P. Gong, G. S. Biging. Comparison of IKONOS and QuickBird imagesfor mapping mangrove species on the Caribbean coast of Panama.Remote Sensing ofEnvironment,2004,91:432-440.
[39]李明阳,申世广,吴翼等.南京紫金山风景林多情境规划方法研究.南京林业大学学报(自然科学版),2007,31(5):29-33
[40]S.E. Franklin, M. A. Wulder, G.R. Gerylo. Texture Analysis of IKONOS Panchromatic Datafor Douglas-fir Forest Age Class Separability in British Columbia.International Journal ofRemote Sensing.2001,22(13):2627—2632.
[41]吴见,彭道黎.基于面向对象的QuickBird影像退耕地树冠信息提取.光谱学与光谱分析,2010,30(9):2533-2536
[42]叶时平,陈超祥,魏玉璋等.基于灰度和纹理特征的QuickBird影像中土地利用信息的提取.通信学报,2008,29(8):129-135
[43]Mena J B, Malpica J A. An automatic method for road extraction in rural and semi-urban arcasstarting from high resolution satellite imagery.Pattern Recognition Letters,2005,26:1201-1220.
[44]翁永玲,范兴旺,胡伍生等.基于QuickBird数据的输电线路径优选中地表信息提取.东南大学学报(自然科学版),2010,40(3):587-592
[45]张海涛,贾光军,虞欣.基于GeoEye-1卫星影像的立体测图技术研究.测绘通报,2010(12):43-46
[46]张治清,何宗.GeoEye-1多光谱与全色影像融合的适应性及质量评价研究.西南师范大学学报(自然科学版),2011,36(1):203-208
[47]虞欣,李和军,贾光军等.GeoEye-1卫星影像定向精度初步分析.测绘通报,2011(1):28-30
[48]黄慧萍,吴炳方.地物大小、对象尺度、影像分辨率的关系分析.遥感技术与应用,2006,21(3):243-248
[49]杜永明,秦其明.不同分辨率对遥感影像中识别人造地物的影响.遥感技术与应用,2001,16(4):214-217
[50]苏理宏,李小文,黄裕霞.遥感尺度问题研究进展.地球科学进展,2001,16(4):544-548
[51]柴渊,李万东.土地利用动态遥感监测技术与方法.地质出版社,2011
[52]李海洋.遥感图像分类方法综述.林业科技情报,2008,40(1):4-5
[53]刘瑞祯,于仕琪.OpenCV教程基础篇.北京航空航天大学出版社,2007
[54]卫亚星,王莉雯.遥感图像增强方法分析.测绘空间地理信息,2006,29(2):3-7
[55]http://www.opencv.org.cn/index.php/VC_2008_Express%E4%B8%8B%E5%AE%89%E8%A3%85OpenCV2.0/2.1