基于CB数学形态学的边缘检测技术研究
|
摘要
边缘检测一直是图像处理领域研究比较多的课题,也是一个热点课题。经过多年的研究,许多算法被提出。边缘检测算法主要分为以下六大类:空域微分算子法、小波变换法、神经网络法、分形理论法、模糊检测法、数学形态学法。其中数学形态学作为一种非线性滤波算子被广泛地研究,主要方法有:形态学梯度法、Top-hat法和多尺度法。在经典形态学中,各种运算均以结构元素与图像形态上的匹配程度来决定变换的结果。因此当需要处理的对象具有不同的表现形态时,不同的结构元素就必须被同时使用。这导致了很大的结构元素集,不仅使系统效率低,也难以获得满意的效果。基于这种认识,产生了基于轮廓结构元素的形态学(Contour Bougie),简称为CB形态学。这种形态学运算建立在新的处理原则上,从而弱化了结构元素形状对处理结果的影响。作为一种新理论,CB形态学在滤除噪声以及保护图像细节信息方面有着无可比拟的优势。本文主要讨论CB形态学的相关知识,并在其基本运算的基础上,引入多尺度的概念,进而提出了基于CB形态学的多尺度边缘检测新算法。
本文给出了经典形态学的多尺度算法、CB形态学的单尺度算法和CB形态学的多尺度的算法,并从处理结果和计算量等多个角度进行了比较,讨论了各种算法的优点和缺点。
Edge detection of image is always a subject being researched very much, which is also a hot research subject. Many algorithms were put forward during the recent years. There are about six kinds among so many algorithms: Spatial differential edge detector, Wavelet transform, Neural network, Fractal theory, Fuzzy set theory, Mathematical morphology. Mathematical morphology is researched much more and it can centigrade into three kinds: Morphological gradient, Top-hat and Multi-scale. In classical morphology, the results of transformation of all kinds of operation are determined by the matching degree between the structure elements and image. Therefore which when needs to process the object has the different performance shape; the different structure elements must be used. This causes the very big structure elements collection. Not only causes the system efficiency to be low, but also obtains unsatisfactory effect. Based on this kind of understanding, we have produced morphology which based on contour bougie, the abbreviation is the CB Morphology. Its operation based on new processing principle, thus it attenuates the influence of the structure elements. As a new theory, the CB Morphology has incomparable superiority in filtering noise and protecting image detail information. The article mainly discusses the CB morphology, introduces the concept of multi-scale based on its basic operation, and then proposes a new edge detection algorithm which based on CB Morphology.
The article gives the multi-scale algorithm of classical morphology, the single-scale algorithm of CB morphology, the multi-scale algorithm of CB morphology. And so on many aspects have carried on the comparison from the processing result and computation time, and then the advantage and disadvantage of them are discussed.
本文给出了经典形态学的多尺度算法、CB形态学的单尺度算法和CB形态学的多尺度的算法,并从处理结果和计算量等多个角度进行了比较,讨论了各种算法的优点和缺点。
Edge detection of image is always a subject being researched very much, which is also a hot research subject. Many algorithms were put forward during the recent years. There are about six kinds among so many algorithms: Spatial differential edge detector, Wavelet transform, Neural network, Fractal theory, Fuzzy set theory, Mathematical morphology. Mathematical morphology is researched much more and it can centigrade into three kinds: Morphological gradient, Top-hat and Multi-scale. In classical morphology, the results of transformation of all kinds of operation are determined by the matching degree between the structure elements and image. Therefore which when needs to process the object has the different performance shape; the different structure elements must be used. This causes the very big structure elements collection. Not only causes the system efficiency to be low, but also obtains unsatisfactory effect. Based on this kind of understanding, we have produced morphology which based on contour bougie, the abbreviation is the CB Morphology. Its operation based on new processing principle, thus it attenuates the influence of the structure elements. As a new theory, the CB Morphology has incomparable superiority in filtering noise and protecting image detail information. The article mainly discusses the CB morphology, introduces the concept of multi-scale based on its basic operation, and then proposes a new edge detection algorithm which based on CB Morphology.
The article gives the multi-scale algorithm of classical morphology, the single-scale algorithm of CB morphology, the multi-scale algorithm of CB morphology. And so on many aspects have carried on the comparison from the processing result and computation time, and then the advantage and disadvantage of them are discussed.
引文
[1]Rafael C.Gonzalez,Richard E.Woods,Steven L.Eddins.Digital Image Processing Using MATLAB.北京.电子工业出版社.2004.5.334-375.
[2]孙即祥.图像分析.北京.科学出版社.2005.179-207.
[3]龚炜,石青云,程民德.数字空间中数学形态学理论及应用.北京.科学出版社.1997.426-454.
[4]章毓晋.图像工程-图像处理和分析.清华大学出版社.1999.
[5]孙祥,黄晓鸣.基于MATLAB的集成化图像处理系统.科学技术与工程.2007.Vol 7.5386-5388
[6]BaiXiangzhi,ZhouFugen.Edge Detection Based on Mathematical and Iterative Thresholding.Computational Intelligence and Security 2006 International Conference on Volume 2.2006.1849-1852.
[7]B.Chanda,K.K.Malay,Y.V.Padmaja.A multiscale morphological filters.Patteren R.ecognition.Vol.31.1998.1469-1478
[8]DaweiQi,FanGuo,LeiYu.Medical Image Edge Detection Based on Omni-directional Multi-Scale Structure Element of Mathematical Morphology.International Conference on Automation and Logistics August 18-21.2007.Jinan,China..
[9]JiangZetao,ZhaoRongchun.Adaptive Multi-scale Edge Detection Based on Region Energy.Chinese Journal of Aeronautics.Vol 18.2005.237-241.
[10]Zhao Yuqian,GuiWeihua,ChenZhencheng.Medical Image Edge Detection Based on Mathematical Morphology.Engineering in Medicine and Biology 27th Annual Conference September 1-4.2005.Shanghai,China
[11]ZhaoYuqian,GuiWeihua,ChenZhencheng.Edge Detection Based on Multi-structure Elements Morphology.The 6th World Congress on Intelligent Control and Automation.June 21-23.2006.Dalian,China.
[12]WangXiaopeng,LuoJinwen.Edge Detection Based on Regulated Morphological Graient.The 3rd International Conference on Computational Electromagnetics and Its Applications Proceedings.2004
[13]连静,王珂.多尺度形态学图像边缘检测方法.计算机工程与应用.2006.77-79
[14]刘循,游志胜.多尺度形态学图像边缘检测方法.光电工程.2003.Vol 30.56-58
[15]万军,徐汀荣.,基于Laplacian算子的图像边缘检测方法研究.现代电子技术.2004.Vol 21.92-94
[16]魏伟波,芮筱亭.图像边缘检测方法研究.计算机工程与应用.2006.88-91
[17]刑军.基于Sobel算子数字图像的边缘检测.微机发展.2005.Vol 15.48-50
[2]孙即祥.图像分析.北京.科学出版社.2005.179-207.
[3]龚炜,石青云,程民德.数字空间中数学形态学理论及应用.北京.科学出版社.1997.426-454.
[4]章毓晋.图像工程-图像处理和分析.清华大学出版社.1999.
[5]孙祥,黄晓鸣.基于MATLAB的集成化图像处理系统.科学技术与工程.2007.Vol 7.5386-5388
[6]BaiXiangzhi,ZhouFugen.Edge Detection Based on Mathematical and Iterative Thresholding.Computational Intelligence and Security 2006 International Conference on Volume 2.2006.1849-1852.
[7]B.Chanda,K.K.Malay,Y.V.Padmaja.A multiscale morphological filters.Patteren R.ecognition.Vol.31.1998.1469-1478
[8]DaweiQi,FanGuo,LeiYu.Medical Image Edge Detection Based on Omni-directional Multi-Scale Structure Element of Mathematical Morphology.International Conference on Automation and Logistics August 18-21.2007.Jinan,China..
[9]JiangZetao,ZhaoRongchun.Adaptive Multi-scale Edge Detection Based on Region Energy.Chinese Journal of Aeronautics.Vol 18.2005.237-241.
[10]Zhao Yuqian,GuiWeihua,ChenZhencheng.Medical Image Edge Detection Based on Mathematical Morphology.Engineering in Medicine and Biology 27th Annual Conference September 1-4.2005.Shanghai,China
[11]ZhaoYuqian,GuiWeihua,ChenZhencheng.Edge Detection Based on Multi-structure Elements Morphology.The 6th World Congress on Intelligent Control and Automation.June 21-23.2006.Dalian,China.
[12]WangXiaopeng,LuoJinwen.Edge Detection Based on Regulated Morphological Graient.The 3rd International Conference on Computational Electromagnetics and Its Applications Proceedings.2004
[13]连静,王珂.多尺度形态学图像边缘检测方法.计算机工程与应用.2006.77-79
[14]刘循,游志胜.多尺度形态学图像边缘检测方法.光电工程.2003.Vol 30.56-58
[15]万军,徐汀荣.,基于Laplacian算子的图像边缘检测方法研究.现代电子技术.2004.Vol 21.92-94
[16]魏伟波,芮筱亭.图像边缘检测方法研究.计算机工程与应用.2006.88-91
[17]刑军.基于Sobel算子数字图像的边缘检测.微机发展.2005.Vol 15.48-50