机场目标分割与识别方法研究
|
摘要
本论文研究的主要内容为遥感图象中的机场分割与识别。作为一个应用很广泛的领域，目标识别具有很强的针对性和依赖性。在实际应用中，往往是针对不同的目标，提出不同的应用假设，选用不同的算法。即使是同样的目标，在不同的应用假设下，采用的算法也是不同的。面对这种研究现状，本文试图通过对现有目标分割与识别算法的研究，增强对算法的理解，并在前人工作的基础上提出针对低分辨率遥感图象中机场的分割与识别算法。
在目标检测与分割方面，根据象素的不连续性和相似性，把一般图象的分割算法分成四类。本文着重讨论了并行边界分割技术和并行区域分割技术，介绍了其中的一些典型技术并分析其优缺点。最后，针对单个算法的局限性和具体机场目标的特点，作者提出一种由多种分割技术组合而成的机场检测和分割方法。实验验证了方法的有效性。
在目标识别方面，讨论了设计识别系统时需要考虑的若干问题；目标识别复杂度所依赖的因素：讨论了基于形状的特征提取方法。最后，根据机场的结构特征，建立了机场的模型表示，讨论了基于Hough变换的识别方法，实验证明其较好的识别效果。
在充分研究机场检测、分割和识别算法的基础上，本文还讨论了并行程序设计的某些问题，并结合具体的MPP体系结构，对算法实现了并行化。
The thesis is focused on airport target detection, segmentation and recognition for the remote sensing image. As a widely used technology, object recognition has following characteristics: pertinence and dependence. In the actual application, facing the different object, we always propose different supposes and select different methods. Even for the same object, different methods are selected in allusion to different applications. Based on predecessor work, we hope to make some improvements to their research results.
In the aspect of object detection and segmentation, algorithms are classified to four parts by discontinuity and comparability of the pixels. We begin with parallel technology of edge and area segmentation. Then some typical algorithms are introduced. In allusion to the weakness of single algorithm and characteristics of the airport, we propose a method composed of multi-segmentation algorithms. The experiments result show that it works well.
In the aspect of recognition, some questions on design recognition system are discussed. We introduce the factors of recognition complication, the expression of airport model, shape-based feature extraction methods. Afterward, based on characteristics of airport, a recognition algorithm using Hough transformation is introduced and the experiments show it works well.
In the end, on deeper researching airport objects detection, segmentation and recognition, we discuss the parallel program design. Based on some typical MPP hardware structure, we design parallel airport recognition algorithms.
在目标检测与分割方面，根据象素的不连续性和相似性，把一般图象的分割算法分成四类。本文着重讨论了并行边界分割技术和并行区域分割技术，介绍了其中的一些典型技术并分析其优缺点。最后，针对单个算法的局限性和具体机场目标的特点，作者提出一种由多种分割技术组合而成的机场检测和分割方法。实验验证了方法的有效性。
在目标识别方面，讨论了设计识别系统时需要考虑的若干问题；目标识别复杂度所依赖的因素：讨论了基于形状的特征提取方法。最后，根据机场的结构特征，建立了机场的模型表示，讨论了基于Hough变换的识别方法，实验证明其较好的识别效果。
在充分研究机场检测、分割和识别算法的基础上，本文还讨论了并行程序设计的某些问题，并结合具体的MPP体系结构，对算法实现了并行化。
The thesis is focused on airport target detection, segmentation and recognition for the remote sensing image. As a widely used technology, object recognition has following characteristics: pertinence and dependence. In the actual application, facing the different object, we always propose different supposes and select different methods. Even for the same object, different methods are selected in allusion to different applications. Based on predecessor work, we hope to make some improvements to their research results.
In the aspect of object detection and segmentation, algorithms are classified to four parts by discontinuity and comparability of the pixels. We begin with parallel technology of edge and area segmentation. Then some typical algorithms are introduced. In allusion to the weakness of single algorithm and characteristics of the airport, we propose a method composed of multi-segmentation algorithms. The experiments result show that it works well.
In the aspect of recognition, some questions on design recognition system are discussed. We introduce the factors of recognition complication, the expression of airport model, shape-based feature extraction methods. Afterward, based on characteristics of airport, a recognition algorithm using Hough transformation is introduced and the experiments show it works well.
In the end, on deeper researching airport objects detection, segmentation and recognition, we discuss the parallel program design. Based on some typical MPP hardware structure, we design parallel airport recognition algorithms.
引文
[1] Halem N.Contextual, "Image Understanding of Airport Photographs", SPIE, 1981,1521—1532
[2] Huertas A.,"Detect Runways in Complex Airport Scenes",Computer Vision,Graphics and Image Processing, 1990,24(2):43—57
[3] 张会章，郭雷，“一个机场跑道的自动识别系统”，计算机工程，2001．3
[4] 叶斌，彭嘉雄，“基于结构特征的军用机场识别与理解”，华中科技大学学报，2001-3
[5] 刘文萍，陈维军，吴立德，“遥感图象中机场目标的图象分割方法”，红外与毫米波学报，1999．2
[6] 孙斌，“航空图片的自动识别”，浙江大学学报
[7] 孙忠学，荆宝中，“桥梁、机场跑道电视目标图象跟踪与识别方法研究”
[8] 徐胜荣，李忠兴，“自然景物中桥梁目标识别方法的研究”，浙江大学学报，1995．9
[9] 薛峰，王润生，“组合利用统计和结构信息的道路提取算法”，光学学报，2001．4
[10] 王仁生，贾晓光，周建林，“从空间遥感图象的自然背景中提取人造目标的研究”中国图象图形学报，1997．7
[11] 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，2001．12
[12] Murat Tekalp, "Digital Video Processing", Prentice-Hall Press, 1996
[13] 赵荣椿，赵忠明，崔甦生，数字图象处理导论，西北工业大学出版社，1996．7
[14] 章毓晋，图象分割，科学出版社，2001
[15] 贾云得，机器视觉，科学出版社，2000
[16] 王秋让，“基于自动门限化的图象分割及目标提取方法研究”，西北工业大学博士论文
[17] Gonzalez R C, Woods R E. Digital Image. 3rd.ed, Addison-Wesley ,1992
[18] Roberts L G Machine perception of three-dimensional solids. In: Optical and Electro-Optical Information Processing, Tippett J, et al., eds., 1965,159～197
[19] Kirsch R. "Computer determination of the constituent structure of biological images", Computer Biomedical Research, 1971, 4:315～328
[20] Marr D, Hildreth E. "Theory of edge detection", Proceedings of R.Soc.London,1980
[21] Buxton B. "Early Image Processing", Structural Techniques Motivated by Human
Visual Response. University of Surrey, 1984
[22] Canny J. "A computational approach to edge detection", IEEE～PAMI, 1986.8:679～698
[23] Sahoo P K, eh al, A survey of thresholding techniques, CVGIP, 41(2):233～260
[24] 吴一全，朱兆达，“图象处理中阈值选取方法30年(1962～1992)的进展”，数据采集与处理，1993．8
[25] Weszka J S, Rosenfeld A. Histogram modification for the threshold selection. IEEE～SMC, 1979,9:38～72
[26] 章毓晋，图象工程，清华大学出版社，1999
[27] Ridler T W, Calvard S, "Pictures thresholding using an iterative selection method", IEEE～SMC, 1978;8:630～632
[28] Jain A K, Fundamentals of Digital Image Processing, Prentice-Hall, 1989
[29] T.F. Cootes and C.J. Taylor, "Statistical Models of Appearance for Computer Vision", Technical Report, Oct., 2001
[30] MSRA UI group, "Bagging and Boosting", Technical Report, 2002
[31] Robert E. Schapire, "The Boosting Approach to Machine Learning: An Overview", MSRI Workshop on Nonlinear Estimation and Classifieation, 2002
[32] Sam Rowels, "A Unifying Review of Linear Gaussian Models", Technical Report, 2000.
[33] Barry Wilkinson，Michael Allen，并行程序设计，机械工业出版社，2002
[34] 李晓明等，并行计算机体系结构：硬件／软件结合的设计与分析，机械工业出版社，2003
[35] 沈绪榜，MPP嵌入式计算机设计，清华大学出版社，1999
[36] MIDDLE：可兼容的汇编语言，航天科技集团771所内部资料
[2] Huertas A.,"Detect Runways in Complex Airport Scenes",Computer Vision,Graphics and Image Processing, 1990,24(2):43—57
[3] 张会章，郭雷，“一个机场跑道的自动识别系统”，计算机工程，2001．3
[4] 叶斌，彭嘉雄，“基于结构特征的军用机场识别与理解”，华中科技大学学报，2001-3
[5] 刘文萍，陈维军，吴立德，“遥感图象中机场目标的图象分割方法”，红外与毫米波学报，1999．2
[6] 孙斌，“航空图片的自动识别”，浙江大学学报
[7] 孙忠学，荆宝中，“桥梁、机场跑道电视目标图象跟踪与识别方法研究”
[8] 徐胜荣，李忠兴，“自然景物中桥梁目标识别方法的研究”，浙江大学学报，1995．9
[9] 薛峰，王润生，“组合利用统计和结构信息的道路提取算法”，光学学报，2001．4
[10] 王仁生，贾晓光，周建林，“从空间遥感图象的自然背景中提取人造目标的研究”中国图象图形学报，1997．7
[11] 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，2001．12
[12] Murat Tekalp, "Digital Video Processing", Prentice-Hall Press, 1996
[13] 赵荣椿，赵忠明，崔甦生，数字图象处理导论，西北工业大学出版社，1996．7
[14] 章毓晋，图象分割，科学出版社，2001
[15] 贾云得，机器视觉，科学出版社，2000
[16] 王秋让，“基于自动门限化的图象分割及目标提取方法研究”，西北工业大学博士论文
[17] Gonzalez R C, Woods R E. Digital Image. 3rd.ed, Addison-Wesley ,1992
[18] Roberts L G Machine perception of three-dimensional solids. In: Optical and Electro-Optical Information Processing, Tippett J, et al., eds., 1965,159～197
[19] Kirsch R. "Computer determination of the constituent structure of biological images", Computer Biomedical Research, 1971, 4:315～328
[20] Marr D, Hildreth E. "Theory of edge detection", Proceedings of R.Soc.London,1980
[21] Buxton B. "Early Image Processing", Structural Techniques Motivated by Human
Visual Response. University of Surrey, 1984
[22] Canny J. "A computational approach to edge detection", IEEE～PAMI, 1986.8:679～698
[23] Sahoo P K, eh al, A survey of thresholding techniques, CVGIP, 41(2):233～260
[24] 吴一全，朱兆达，“图象处理中阈值选取方法30年(1962～1992)的进展”，数据采集与处理，1993．8
[25] Weszka J S, Rosenfeld A. Histogram modification for the threshold selection. IEEE～SMC, 1979,9:38～72
[26] 章毓晋，图象工程，清华大学出版社，1999
[27] Ridler T W, Calvard S, "Pictures thresholding using an iterative selection method", IEEE～SMC, 1978;8:630～632
[28] Jain A K, Fundamentals of Digital Image Processing, Prentice-Hall, 1989
[29] T.F. Cootes and C.J. Taylor, "Statistical Models of Appearance for Computer Vision", Technical Report, Oct., 2001
[30] MSRA UI group, "Bagging and Boosting", Technical Report, 2002
[31] Robert E. Schapire, "The Boosting Approach to Machine Learning: An Overview", MSRI Workshop on Nonlinear Estimation and Classifieation, 2002
[32] Sam Rowels, "A Unifying Review of Linear Gaussian Models", Technical Report, 2000.
[33] Barry Wilkinson，Michael Allen，并行程序设计，机械工业出版社，2002
[34] 李晓明等，并行计算机体系结构：硬件／软件结合的设计与分析，机械工业出版社，2003
[35] 沈绪榜，MPP嵌入式计算机设计，清华大学出版社，1999
[36] MIDDLE：可兼容的汇编语言，航天科技集团771所内部资料