基于图像区域特征的边缘检测方法研究与实现
|
摘要
在实际图像处理问题中,边缘是图像的最基本特征。边缘检测是图像的压缩、图像的增强、图像的分割以及图像的识别等领域的基础,有着广泛的应用,一直是数字图像处理领域研究的焦点问题和热点问题。
目前在边缘的检测和追踪领域已经提出了许多方法,鉴于所分析的图像特点的差异,不同边缘检测追踪算法的性能具有很大的互异性。尤其,实际处理的图像一般都有噪声点的存在、不封闭曲线和目标对象有公共点,如何保证边缘定位的准确性成为边缘检测需要解决的一个重要问题。因此,需要根据具体应用设计新的边缘检测方法,或者对现有的方法进行改进以得到满意的边缘检测结果。而基于图像区域特征的边缘检测追踪方法逐渐被应用到实际的图像处理中,并且显示出它优于传统方法的特点,针对在相干切片和TO图的断层多边形自动提取交互系统的研发课题,系统需要在具有复杂断层分布和强噪声的相干切片或TO上准确、快速地提取断层多边形轮廓,并且对锯齿较大的多边形进行自动平滑处理,并保存边界参数序列,本文设计了基于图像区域特征的边缘检测算法和系统实施方案,并得到了较好的检测效果。
本文首先研究了经典的图像边缘检测方法,线性滤波技术及其在图像边缘检测中的应用,并通过理论分析以及仿真计算,比较了各种算法在图像边缘检测中的优缺点。其次,研究与图像边缘检测相关的阈值分割二值化图像预处理算法、对传统的边界追踪算法进行分析,与图像区域特征相结合提出一种改进型的八邻域边界跟踪算法；对得到的边缘轮廓进行平滑,对传统的平滑算法的补充和改进,使断层多边形平滑处理达到不失真、高效率。最后,在此基础上,针对项目需求设计了复杂二值图像边界跟踪与提取算法,并将改进的各种算法应用到项目中,设计实现了断层多边形自动提取交互系统。通过实践验证：该系统检测出的断层多边形定位准确、连续、光滑、边缘细节丰富、边界呈单像素宽。该系统在油田地震勘探解释应用中取得良好检测效果,并提高了断层检测的精度和效率。
In image processing, the edge is one of the basic characters of an image. Edge detection has been widely used in image processing, image analysis and computer vision, and it is the basis of them. So, Edge detection is always the hot and focus in the research field of digital image processing.
There have been many algorithms proposed for the edge detection of images. But the existing theories and algorithms for edge detection still have some drawbacks and can not detect edges satisfactorily in some cases. Especially, images obtained from real-world scenes are generally buried in noise, have no closed curves and the target have public points. How to detect edges reliably and accurately in the presence of noise has remained an important issue in the field of edge detection. So, it has been the focus of current research work to find new methods for edge detection with specific application requirements or to make improvements to existing methods. The image edge detection based on regional features have been gradually applied to the actual image processing, and shown that it is superior to the traditional algorithm. This thesis sets about research for need of project, in the process of designing fault polygon automatic extraction interactive system of coherence slice and TO figure, knowing this system require to detect the figure of the Fault Polygon and TO figure quickly and accurately, and get sequences of boundary points'parameters according to need, and deal with the zigzag larger polygon automatically smooth. Considering the complexity fault distribution and noise influence of the coherence slice and TO figure, this thesis first proposes the seismic image edge detection based on regional features and projects to resolve practical problems, and get satisfied result.
Firstly, the classic methods of image edge detection and tracking are introduced, the technique of linear filter and its application is discussed, at the same time, compared with the characteristic of all kinds of methods on image edge detection and tracking by using theoretical analysis and simulated computation. Secondly, Image Preprocessing algorithms of edge detection are researched in this thesis, and images traditional algorithms of edge tracking are analysed, and a improved algorithm for images edge tracking combines with regional features; the traditional smoothing algorithm is complemented and improved, it gets the edge profile smoothly, and makes the processing of fault polygon smoothing lossless and efficient. Finally, on the basis, a boundary tracking and extracting algorithm of the complicated binary images is presented according to need, and a project applying all kinds of algorithm synthetically is designed for fault polygon automatic extraction interactive system. And the result indicates that the detected single-point edge has no fake edge, smooth and continuous, at the same time it has more abundant details of edges and has good localization. This system improves the efficiency and accuracy of fault interpretation and makes the coherence technique play a important role in 3D data interpretation. The practical application showed that coherence slice fault polygon detecting system is a very efficient tool for the whole 3D data interpretation and it reduces the cycle of 3D data interpretation.
目前在边缘的检测和追踪领域已经提出了许多方法,鉴于所分析的图像特点的差异,不同边缘检测追踪算法的性能具有很大的互异性。尤其,实际处理的图像一般都有噪声点的存在、不封闭曲线和目标对象有公共点,如何保证边缘定位的准确性成为边缘检测需要解决的一个重要问题。因此,需要根据具体应用设计新的边缘检测方法,或者对现有的方法进行改进以得到满意的边缘检测结果。而基于图像区域特征的边缘检测追踪方法逐渐被应用到实际的图像处理中,并且显示出它优于传统方法的特点,针对在相干切片和TO图的断层多边形自动提取交互系统的研发课题,系统需要在具有复杂断层分布和强噪声的相干切片或TO上准确、快速地提取断层多边形轮廓,并且对锯齿较大的多边形进行自动平滑处理,并保存边界参数序列,本文设计了基于图像区域特征的边缘检测算法和系统实施方案,并得到了较好的检测效果。
本文首先研究了经典的图像边缘检测方法,线性滤波技术及其在图像边缘检测中的应用,并通过理论分析以及仿真计算,比较了各种算法在图像边缘检测中的优缺点。其次,研究与图像边缘检测相关的阈值分割二值化图像预处理算法、对传统的边界追踪算法进行分析,与图像区域特征相结合提出一种改进型的八邻域边界跟踪算法；对得到的边缘轮廓进行平滑,对传统的平滑算法的补充和改进,使断层多边形平滑处理达到不失真、高效率。最后,在此基础上,针对项目需求设计了复杂二值图像边界跟踪与提取算法,并将改进的各种算法应用到项目中,设计实现了断层多边形自动提取交互系统。通过实践验证：该系统检测出的断层多边形定位准确、连续、光滑、边缘细节丰富、边界呈单像素宽。该系统在油田地震勘探解释应用中取得良好检测效果,并提高了断层检测的精度和效率。
In image processing, the edge is one of the basic characters of an image. Edge detection has been widely used in image processing, image analysis and computer vision, and it is the basis of them. So, Edge detection is always the hot and focus in the research field of digital image processing.
There have been many algorithms proposed for the edge detection of images. But the existing theories and algorithms for edge detection still have some drawbacks and can not detect edges satisfactorily in some cases. Especially, images obtained from real-world scenes are generally buried in noise, have no closed curves and the target have public points. How to detect edges reliably and accurately in the presence of noise has remained an important issue in the field of edge detection. So, it has been the focus of current research work to find new methods for edge detection with specific application requirements or to make improvements to existing methods. The image edge detection based on regional features have been gradually applied to the actual image processing, and shown that it is superior to the traditional algorithm. This thesis sets about research for need of project, in the process of designing fault polygon automatic extraction interactive system of coherence slice and TO figure, knowing this system require to detect the figure of the Fault Polygon and TO figure quickly and accurately, and get sequences of boundary points'parameters according to need, and deal with the zigzag larger polygon automatically smooth. Considering the complexity fault distribution and noise influence of the coherence slice and TO figure, this thesis first proposes the seismic image edge detection based on regional features and projects to resolve practical problems, and get satisfied result.
Firstly, the classic methods of image edge detection and tracking are introduced, the technique of linear filter and its application is discussed, at the same time, compared with the characteristic of all kinds of methods on image edge detection and tracking by using theoretical analysis and simulated computation. Secondly, Image Preprocessing algorithms of edge detection are researched in this thesis, and images traditional algorithms of edge tracking are analysed, and a improved algorithm for images edge tracking combines with regional features; the traditional smoothing algorithm is complemented and improved, it gets the edge profile smoothly, and makes the processing of fault polygon smoothing lossless and efficient. Finally, on the basis, a boundary tracking and extracting algorithm of the complicated binary images is presented according to need, and a project applying all kinds of algorithm synthetically is designed for fault polygon automatic extraction interactive system. And the result indicates that the detected single-point edge has no fake edge, smooth and continuous, at the same time it has more abundant details of edges and has good localization. This system improves the efficiency and accuracy of fault interpretation and makes the coherence technique play a important role in 3D data interpretation. The practical application showed that coherence slice fault polygon detecting system is a very efficient tool for the whole 3D data interpretation and it reduces the cycle of 3D data interpretation.
引文
[1]李弼程,彭天强,彭波,等.智能图像处理技术.北京：电子工业出版社,2004.
[2]高彦平.图像增强方法的研究与实现[D].济南：山东科技大学,2005.
[3]J. S. Lee, Digital Enhancement and Noising Filtering by Using of Local Statistics [J],IEEE Transactions on Pattern Analysis and Machine Intelligence,1980, VOL.PAMI-2:165-168.
[4]A.Gupta,R.Jain, Visual information retrieval[C], communications of the ACM,1997, 40(5):70-79.
[5]M.Flickner,H.Sawhney W.Niblaek,etal.Query by image and video content:The QBIC system[J]. IEEE ComPuter,1995,28(9):23-32.
[6]T.Kawaguchi, M.Rizon.Iris detection using intensity and edge information.Pattern Recognition,2003, (36):549-562
[7]M.A.Kbir, H.Benkirane, K.Maalmi, R.Benslimane, Hierarchical fuzzy partition for pattern classification with fuzzy if-then rules[J], Pattern Recognition Lettem,2000,21(10):503-509
[8]卢洋.分数阶微分应用于图像边缘检测的研究[D].大连交通大学,2009.
[9]张小磊.图像边缘处理技术方法的研究及应用[D].南京信息工程大学,2009.
[10]陈初侠.图像滤波及边缘检测与增强技术研究[D].合肥工业大学,2009.
[11]段瑞玲,李庆祥,李玉和.图像边缘检测方法研究综述[J].光学技术,2005,31(3)：415-1419.
[12]曾庆贵.特定材质的物体轮廓提取技术研究[D].山东建筑大学,2008.
[13]姚峰林,詹海英,李元宗.机器视觉中的边缘检测技术研究[J].机械工程与自动化,2005,3(1)：108-110.
[14]马蓓.车型识别技术在视频监控中的应用[D].西安电子科技大学,2010.
[15]施达雅.基于噪声统计模型的区域分割[D].福建师范大学,2009.
[16]M.F.Swainand,D.H.ballard.Colorindexing[J].IntI.J.Computer.Vis.1991,7(1),11-32.
[17]M.Stricker,and M.Orengo."Similaryity of color images,"SPIEStorage and Retrieval for Image and Video Data basesⅢ.vol.2185.pp.381-392,Feb.1995.
[18]胡广寰.基于内容图像检索中图像语义分类技术研究[D].浙江大学,2005.
[19]Jing Huang,S Ravi Kumary,Mandar Mitraz,Wei-Jing,Zhux Ramin Zabih.Image Indexing Using Color Correlograms[C].Proceeding on the IEEE Computer Society Conference on Computer Vision and Pattern Recognition,1997,3(3):762-768.
[20]Tamura H,Mori S and Yamawaki T.Texture Features Corresponding to Visual Perception[J].IEEE Trans On Systems,Man and Cybernetics,1978,8(6)460-473.
[21]Jianchang Mao, Anil K Jain. Texture Classification and Segmentation Using Multiresolution Simultaneous Autoregressive Models[J]. Pattern Recognition,1992, 25(2):173-188.
[22]胡海波.基于图像测量技术CDT荫罩板检测系统研究[D].南京理工大学,2005.
[23]L. G Robert. Machine PercePtion of three-dimensional solids. In:OPtieal and Eleetro-QPtieal Information Proeessing, TIPPett J, et al. eds,1965:159-197.
[24]I. Sobel, Camera Models and Machine PercePtion StanfordAl,1970, Memo121.
[25]J. M. S. Prewitt. Object enhancement and extraetion in B. S. LIPkin and A Rosenfeldeds. Pieture Proeessing and PsyehoPietories,1970.
[26]Castleman K R. Digital Image Processing[M]. Tsinghua University Publishing House,1998
[27]Pratt W R. Digital image processing[M].Academic Press,1977
[28]叶枫.基于灰度图像的边缘检测方法研究[D].西北工业大学,2003.
[29]Fangwei Zhao,etal. Use of the Laplacian of Gaussian operator in prostate ultrasound image processing [J], Proceedings of the 20th Annual International Conference of the IEEE[J],1998,2:812-815
[30]Marr D. Theory of edge detection. Proc Roy Soc,1980,B(27):187-217
[31]Canny J F.A computational approach to edge detection[J].IEEE Trans on PAMI,1986,8(6):679-698
[32]冯永强.图像边缘信息处理技术研究及应用[D].大庆石油学院,2007.
[33]张涛,于飞,杨树勋.基于数字图像处理的油管螺纹参数的智能检测[J].石油工业计算机应用,2005(4).
[34]崔凤魁,张丰收,白露,等.二值图像目标邻域点法边界跟踪算法[J].洛阳：洛阳工学院学报,2001,22(1)：28-34.
[35]李云,鲍苏苏,杨璐.二值图象中物体区域的选定及外边缘跟踪技术[J].广东：华南师范大学学报,2008(3)：27-29.
[36]吴丹.一种快速准确的细化算法[J].计算机与现代化,2003(1)：6-7.
[37]张强.基于几何特征的目标识别及跟踪技术的研究[D].长春理工,2008.
[38]王福生,齐国清.二值图像中目标物体轮廓的边界跟踪算法[J].大连：大连海事大学学报,2006,32(1)：62-64.
[39]王听,曲波,王厚枢.一种不规则封闭形状的轮廓跟踪算法[J].南京：南京航空航天大学学报,1996,28(4)：580-583.
[40]柳稼航,杨建峰,单新建,等.一种基于优先搜索方向的边界跟踪算法[J].兰州：遥感技术与应用,2004,19(3)：209-213.
[41]刘相滨,向坚持,阳波.基于八邻域边界跟踪的标号算法[J].北京：计算机工程与应用,2001,23：125-126.
[42]张丰收.金相组织自动识别及定量分析系统[D].江苏理工,2001.
[43]王宗彦,梁远蕾,李奇敏等.断层数据三维重构技术的研究进展[J].北京：工程图学学报,2002(1).
[44]罗斌,汪炳权.三维表面平滑约束断层图象形状插值[J].北京：中国图象图形学报,1996,1(4).
[45]刘彦花,叶国华.矢量曲线抽稀算法分析[J].武汉：城市勘测,2001(4).
[46]易辉伟.基于AUTO CAD的自动跟踪扫描矢量化系统的设计[D].中南大学,2004.
[47]高凯.立木树干弯曲度自动识别方法的研究[D].北京林业大学,2004.
[48]孙仁山.立木枝干机器视觉识别技术研究[D].北京林业大学,2006.
[49]杜婧,张献州.道格拉斯-普克算法的改进及其在管线制图中的应用[J].上海：铁道勘察,2008(4)
[50]张向君,李幼铭,钟吉太,等.三维相干切片断层多边形检测[J].成都：物探化探计算技术,2001,23(4)：295-298.
[51]周维.面向过程和面向对象分析方法对比研究[D].华中科技大学,2002.
[52]王兆湖,刘财,王建民,张向君,王允清,唐世伟.基于层位解释的断层多边形自动组合技术[J].石油地球物理勘探,2009(1)：106.
[2]高彦平.图像增强方法的研究与实现[D].济南：山东科技大学,2005.
[3]J. S. Lee, Digital Enhancement and Noising Filtering by Using of Local Statistics [J],IEEE Transactions on Pattern Analysis and Machine Intelligence,1980, VOL.PAMI-2:165-168.
[4]A.Gupta,R.Jain, Visual information retrieval[C], communications of the ACM,1997, 40(5):70-79.
[5]M.Flickner,H.Sawhney W.Niblaek,etal.Query by image and video content:The QBIC system[J]. IEEE ComPuter,1995,28(9):23-32.
[6]T.Kawaguchi, M.Rizon.Iris detection using intensity and edge information.Pattern Recognition,2003, (36):549-562
[7]M.A.Kbir, H.Benkirane, K.Maalmi, R.Benslimane, Hierarchical fuzzy partition for pattern classification with fuzzy if-then rules[J], Pattern Recognition Lettem,2000,21(10):503-509
[8]卢洋.分数阶微分应用于图像边缘检测的研究[D].大连交通大学,2009.
[9]张小磊.图像边缘处理技术方法的研究及应用[D].南京信息工程大学,2009.
[10]陈初侠.图像滤波及边缘检测与增强技术研究[D].合肥工业大学,2009.
[11]段瑞玲,李庆祥,李玉和.图像边缘检测方法研究综述[J].光学技术,2005,31(3)：415-1419.
[12]曾庆贵.特定材质的物体轮廓提取技术研究[D].山东建筑大学,2008.
[13]姚峰林,詹海英,李元宗.机器视觉中的边缘检测技术研究[J].机械工程与自动化,2005,3(1)：108-110.
[14]马蓓.车型识别技术在视频监控中的应用[D].西安电子科技大学,2010.
[15]施达雅.基于噪声统计模型的区域分割[D].福建师范大学,2009.
[16]M.F.Swainand,D.H.ballard.Colorindexing[J].IntI.J.Computer.Vis.1991,7(1),11-32.
[17]M.Stricker,and M.Orengo."Similaryity of color images,"SPIEStorage and Retrieval for Image and Video Data basesⅢ.vol.2185.pp.381-392,Feb.1995.
[18]胡广寰.基于内容图像检索中图像语义分类技术研究[D].浙江大学,2005.
[19]Jing Huang,S Ravi Kumary,Mandar Mitraz,Wei-Jing,Zhux Ramin Zabih.Image Indexing Using Color Correlograms[C].Proceeding on the IEEE Computer Society Conference on Computer Vision and Pattern Recognition,1997,3(3):762-768.
[20]Tamura H,Mori S and Yamawaki T.Texture Features Corresponding to Visual Perception[J].IEEE Trans On Systems,Man and Cybernetics,1978,8(6)460-473.
[21]Jianchang Mao, Anil K Jain. Texture Classification and Segmentation Using Multiresolution Simultaneous Autoregressive Models[J]. Pattern Recognition,1992, 25(2):173-188.
[22]胡海波.基于图像测量技术CDT荫罩板检测系统研究[D].南京理工大学,2005.
[23]L. G Robert. Machine PercePtion of three-dimensional solids. In:OPtieal and Eleetro-QPtieal Information Proeessing, TIPPett J, et al. eds,1965:159-197.
[24]I. Sobel, Camera Models and Machine PercePtion StanfordAl,1970, Memo121.
[25]J. M. S. Prewitt. Object enhancement and extraetion in B. S. LIPkin and A Rosenfeldeds. Pieture Proeessing and PsyehoPietories,1970.
[26]Castleman K R. Digital Image Processing[M]. Tsinghua University Publishing House,1998
[27]Pratt W R. Digital image processing[M].Academic Press,1977
[28]叶枫.基于灰度图像的边缘检测方法研究[D].西北工业大学,2003.
[29]Fangwei Zhao,etal. Use of the Laplacian of Gaussian operator in prostate ultrasound image processing [J], Proceedings of the 20th Annual International Conference of the IEEE[J],1998,2:812-815
[30]Marr D. Theory of edge detection. Proc Roy Soc,1980,B(27):187-217
[31]Canny J F.A computational approach to edge detection[J].IEEE Trans on PAMI,1986,8(6):679-698
[32]冯永强.图像边缘信息处理技术研究及应用[D].大庆石油学院,2007.
[33]张涛,于飞,杨树勋.基于数字图像处理的油管螺纹参数的智能检测[J].石油工业计算机应用,2005(4).
[34]崔凤魁,张丰收,白露,等.二值图像目标邻域点法边界跟踪算法[J].洛阳：洛阳工学院学报,2001,22(1)：28-34.
[35]李云,鲍苏苏,杨璐.二值图象中物体区域的选定及外边缘跟踪技术[J].广东：华南师范大学学报,2008(3)：27-29.
[36]吴丹.一种快速准确的细化算法[J].计算机与现代化,2003(1)：6-7.
[37]张强.基于几何特征的目标识别及跟踪技术的研究[D].长春理工,2008.
[38]王福生,齐国清.二值图像中目标物体轮廓的边界跟踪算法[J].大连：大连海事大学学报,2006,32(1)：62-64.
[39]王听,曲波,王厚枢.一种不规则封闭形状的轮廓跟踪算法[J].南京：南京航空航天大学学报,1996,28(4)：580-583.
[40]柳稼航,杨建峰,单新建,等.一种基于优先搜索方向的边界跟踪算法[J].兰州：遥感技术与应用,2004,19(3)：209-213.
[41]刘相滨,向坚持,阳波.基于八邻域边界跟踪的标号算法[J].北京：计算机工程与应用,2001,23：125-126.
[42]张丰收.金相组织自动识别及定量分析系统[D].江苏理工,2001.
[43]王宗彦,梁远蕾,李奇敏等.断层数据三维重构技术的研究进展[J].北京：工程图学学报,2002(1).
[44]罗斌,汪炳权.三维表面平滑约束断层图象形状插值[J].北京：中国图象图形学报,1996,1(4).
[45]刘彦花,叶国华.矢量曲线抽稀算法分析[J].武汉：城市勘测,2001(4).
[46]易辉伟.基于AUTO CAD的自动跟踪扫描矢量化系统的设计[D].中南大学,2004.
[47]高凯.立木树干弯曲度自动识别方法的研究[D].北京林业大学,2004.
[48]孙仁山.立木枝干机器视觉识别技术研究[D].北京林业大学,2006.
[49]杜婧,张献州.道格拉斯-普克算法的改进及其在管线制图中的应用[J].上海：铁道勘察,2008(4)
[50]张向君,李幼铭,钟吉太,等.三维相干切片断层多边形检测[J].成都：物探化探计算技术,2001,23(4)：295-298.
[51]周维.面向过程和面向对象分析方法对比研究[D].华中科技大学,2002.
[52]王兆湖,刘财,王建民,张向君,王允清,唐世伟.基于层位解释的断层多边形自动组合技术[J].石油地球物理勘探,2009(1)：106.