图像边缘检测算法-英文文献-翻译-中英文翻译

图像边缘检测算法-英文文献- 翻译 阿房宫赋翻译下载德汉翻译pdf阿房宫赋翻译下载阿房宫赋翻译下载翻译理论.doc -中英文翻译 青岛大学毕业 设计 领导形象设计圆作业设计ao工艺污水处理厂设计附属工程施工组织设计清扫机器人结构设计 英文资料翻译 image edge examination algorithm Abstract Digital image processing took a relative quite young discipline, is following the computer technology rapid development, day by day obtains the widespread application.The edge took the image one kind of basic characteristic, in the pattern recognition, the image division, the image intensification as well as the image compression and so on in the domain has a more widespread application.Image edge detection method many and varied, in which based on brightness algorithm, is studies the time to be most long, the theory develops the maturest method, it mainly is through some difference operator, calculates its gradient based on image brightness the change, thus examines the edge, mainly has Robert, Laplacian, Sobel, Canny, operators and so on LOG。 First as a whole introduced digital image processing and the edge detection survey, has enumerated several kind of at present commonly used edge detection technology and the algorithm, and selects two kinds to use Visual the C language programming realization, through withdraws the image result to two algorithms the comparison, the research discusses their good and bad points. Foreword In image processing, as a basic characteristic, the edge of theimage, which is widely used in the recognition, segmentation,intensification and compress of the image, is often applied tohigh-level domain.There are many kinds of ways to detect the edge. Anyway, there aretwo main techniques: one is classic method based on the gray grade ofevery pixel; the other one is based on wavelet and its multi-scalecharacteristic. The first method, which is got the longest research,get the edge according to the variety of the pixel gray. The maintechniques are Robert, Laplace, Sobel, Canny and LOG algorithm. The second method, which is based on wavelet transform, utilizes theLipschitz exponent characterization of the noise and singular signaland then achieve the goal of removing noise and distilling the realedge lines. In recent years, a new kind of detection method, which basedon the phase information of the pixel, is developed. We need hypothesizenothing about images in advance. The edge is easy to find in frequencydomain. It’s a reliable method. In chapter one, we give an overview of the image edge. And inchapter two, some classic detection algorithms are introduced. Thecause of positional error is analyzed, and then discussed a moreprecision method in edge orientation. In chapter three, wavelet theoryis introduced. The detection methods based on sampling wavelettransform, which can extract maim edge of the image effectively, andnon-sampling wavelet transform, which can remain the optimum spatialinformation, are recommended respectively. In the last chapter of thisthesis, the algorithm based on phase information is introduced. Usingthe log Gabor wavelet, two-dimension filter is constructed, many kindsof edges are detected, including Mach Band, which indicates it is aoutstanding and bio-simulation method。May all the work in this paper is of some value to research and applications of image edge detection. First chapter introduction ?1.1 image edge examination introduction The image edge is one of image most basic characteristics, often is carrying image majority of informations。But the edge exists in the image irregular structure and in not the steady phenomenon, also namely exists in the signal point of discontinuity place, these spots have given the image outline 第1页 青岛大学毕业设计英文资料翻译 position, these outlines are frequently we when the imagery processing needs the extremely important some representative condition, this needs us to examine and to withdraw its edge to an image。But the edge examination algorithm is in the imagery processing question one of classical technical difficult problems, its solution carries on the high level regarding us the characteristic description, the recognition and the understanding and so on has the significant influence; Also because the edge examination all has in many aspects the extremely important use value, therefore how the people are devoting continuously in study and solve the structure to leave have the good nature and the good effect edge examination operator question。In the usual situation, we may the signal in singular point and the point of discontinuity thought is in the image peripheral point, its nearby gradation change situation may reflect from its neighboring picture element gradation distribution gradient。 According to this characteristic, we proposed many kinds of edge examination operator: If Robert operator, Sobel operator, Prewitt operator, Laplace operator and so on.These methods many are wait for the processing picture element to carry on the gradation analysis for the central neighborhood achievement the foundation, realized and has already obtained the good processing effect to the image edge extraction.。But this kind of method simultaneously also exists has the edge picture element width, the noise jamming is serious and so on the shortcomings, even if uses some auxiliary methods to perform the denoising, also corresponding can bring the flaw which the edge fuzzy and so on overcomes with difficulty。Along with the wavelet analysis appearance， its good time frequency partial characteristic by the widespread application in the imagery processing and in the pattern recognition domain, becomes in the signal processing the commonly used method and the powerful tool。Through the wavelet analysis, may interweave decomposes in the same place each kind of composite signal the different frequency the block signal， but carries on the edge examination through the wavelet transformation, may use its multi-criteria and the multi-resolution nature fully， real effective expresses the image the edge characteristic。When the wavelet transformation criterion reduces, is more sensitive to the image detail; But when the criterion increases, the image detail is filtered out, the examination edge will be only the thick outline.This characteristic is extremely useful in the pattern recognition, we may be called this thick outline the image the main edge.If will be able an image main edge clear integrity extraction, this to the goal division, the recognition and so on following processing to bring the enormous convenience.Generally speaking, the above method all is the work which does based on the image luminance information。 In the multitudinous scientific research worker under, has obtained the very good effect diligently.But, because the image edge receives physical condition and so on the illumination influences quite to be big above, often enables many to have a common shortcoming based on brightness edge detection method, that is the edge is not continual, does not seal up.Considered the phase information in the image importance as well as its stable characteristic, causes using the phase information to carry on the imagery processing into new research topic。 In this paper soon introduces one 第2页 青岛大学毕业设计英文资料翻译 kind based on the phase image characteristic examination method - - phase uniform method.It is not uses the image the luminance information, but is its phase characteristic, namely supposition image Fourier component phase most consistent spot achievement characteristic point.Not only it can examine brightness characteristics and so on step characteristic, line characteristic, moreover can examine Mach belt phenomenon which produces as a result of the human vision sensation characteristic.Because the phase uniformity does not need to carry on any supposition to the image characteristic type, therefore it has the very strong versatility。 ?1.2 image edge definition The image majority main information all exists in the image edge, the main performance for the image partial characteristic discontinuity, is in the image the gradation change quite fierce place, also is the signal which we usually said has the strange change place。 The strange signal the gradation change which moves towards along the edge is fierce, usually we divide the edge for the step shape and the roof shape two kind of types (as 1).In the step edge two side grey levels have the obvious change; But shown in Figure 1- the roof shape edge is located the gradation increase and the reduced intersection point.May portray the peripheral point in mathematics using the gradation derivative the change, to the step edge, the roof shape edge asks its step, the second time derivative separately。 To an edge, has the possibility simultaneously to have the step and the line edge characteristic. For example on a surface, changes from a plane to the normal direction different another plane can produce the step edge; If this surface has the edges and corners which the regular reflection characteristic also two planes form quite to be smooth, then works as when edges and corners smooth surface normal after mirror surface reflection angle, as a result of the regular reflection component, can produce the bright light strip on the edges and corners smooth surface, such edge looked like has likely superimposed a line edge in the step edge. Because edge possible and in scene object important characteristic correspondence, therefore it is the very important image characteristic。For instance, an object outline usually produces the step edge, because the object image intensity is different with the background image intensity。 ?1.3 paper selected topic theory significance The paper selected topic originates in holds the important status and the function practical application topic in the image project.The so-called image project discipline is refers foundation discipline and so on mathematics, optics principles, the discipline which in the image application unifies which accumulates the technical background develops.The image project content is extremely rich, and so on divides into three levels differently according to the abstract degree and the research technique: Imagery processing, image analysis and image understanding。As shown in Figure 1-2, in the chart, the image division is in between the image analysis and the imagery processing, its meaning is, the image division is from the imagery processing to the image analysis essential step, also is further understands the image the foundation。 The image division has the important influence to the characteristic.The image division and based on the division goal expression, the characteristic extraction and the parameter survey and so on transforms the primitive image as a more abstract more compact form, causes the high-level image analysis and possibly understands into.But the edge examination is the image division core content, therefore the edge examination holds the important status and the function in the image project.Therefore the edge examination 第3页 青岛大学毕业设计英文资料翻译 research always is in the image engineering research the hot spot and the focal point, moreover the people enhance unceasingly to its attention and the investment。 ?1.4 this article prime tasks Algorithm content ?1.4.1 Introduced and has analyzed the classics image edge examination algorithm, summarized each algorithm good and bad points, has given the image edge examination result, and emphatically take the LOG algorithm as the example, embarked from the noise and the edge shape viewpoint has analyzed the reason which the edge position error produced; Introduced in one kind of LOG algorithm the quite precise definite edge method。 ?1.4.2 Wavelet theory Has studied the wavelet elementary theory, summarized the signal as well as the noise Lip index nature, and in based on in the non-sampling wavelet transformation image characteristic extraction algorithm foundation, unifies the auto-adapted denoising method, has made certain improvement to this method, obtained the quite satisfactory effect, denoising ability had the quite big enhancement; Then introduced one kind based on the sampling wavelet examination image main edge method。 ?1.4.3 Novel algorithm The system has studied one kind quite novel based on the phase image characteristic extraction algorithm - - phase uniform algorithm, and has given its simple algorithm.Has given in the unidimensional situation algorithm simulation step, analyzed expanded to the two-dimensional method, and explained by the edge examination result the phase uniform algorithm conformed to the human vision characteristic。 ?1.5 this article content arrangement In the first chapter, the author has given an outline explanation to the image edge examination, and explained carries on the image edge examination the vital significance.In second chapter, the system introduced the quite classical image edge examination operator and the concrete realization principle, have analyzed each algorithm existence insufficiency by the edge examination result.Finally, from the noise influence and edge shape obtaining, take the LOG algorithm as the example, has analyzed the reason which the false edge as well as edge shifting produces.Finally introduced in one kind of LOG algorithm the quite precise definite edge method. In third chapter, the author system introduced the present quite popular wavelet theory, and introduced emphatically the multi-criterion concept and the signal Lip index, and by the noise and the signal Lip index characteristic, carries on the extraction in the non-sampling wavelet transformation foundation to the image edge.In order to strengthen the edge image anti-chirp ability, but also the algorithm has made certain improvement regarding this, the auto-adapted denoising method will use in the edge detection, has obtained the satisfying effect.Finally also introduces one kind based on the sampling wavelet examination image main edge method.In the this article fourth chapter, introduced one kind quite novel based on the phase image characteristic extraction algorithm - - phase uniform algorithm.From unidimensional algorithm introduction obtaining, has given under the unidimensional signal simulation result, and expands gradually two-dimensionally.Explained through the simulation result this algorithm robustness quite is strong, moreover conforms to humanity's visual system performance. Second chapter classical image edge examination algorithm This chapter first simply introduced a classics step edge examination essential method in 2.1.Then 2.2 and 2.3 distinctions elaborated specifically the classical derivative 第4页 青岛大学毕业设计英文资料翻译 operator and the linear filtering operator realization method, and has given each algorithm result comparison in 2.4.In 2.5, compared with the concrete analysis noise and the edge shape the reason which produced to the edge pointing accuracy influence as well as the false edge, and has given in the unidimensional situation simulation result, has drawn the conclusion.In 2.6, the image positive and negative edge which picks out using the LOG algorithm, compared with the precise localization image real edge, the final output was two value single picture element image. ?2.1 classical edge examination essential method We knew that, the edge examination essence is uses some algorithm to withdraw in the image the object and the background junction demarcation line.We define the edge for the image in the gradation occur the rapid change region boundary.The image gradation change situation may use the image gradation distribution the gradient to reflect, therefore we may use the partial image differential technology to obtain the edge examination operator. The edge examination algorithm has the following four steps ( its process as shown in Figure2-1): Filter: The edge examination algorithm mainly is based on an image intensity step and the second time derivative, but the derivative computation is very sensitive to the noise, therefore must use the filter to improve and the noise related edge detector performance.Needs to point out that, the majority filter have also caused the edge intensity loss while noise reduction, therefore, strengthens the edge and between the noise reduction needs compromised. Enhancement: Strengthens the edge the foundation is determines the image each neighborhood intensity the change value.The enhancement algorithm may (or partial) the intensity value has the neighborhood the remarkable change spot to reveal suddenly.The edge strengthens is generally completes through the computation gradient peak-to-peak value. Examination: Has many point gradient peak-to-peak value in the image quite to be big, but these in the specific application domain not all is the edge, therefore should use some method to determine which select is the peripheral points.The simple edge examination criterion is the gradient peak-to-peak value threshold value criterion. Localization: If some application situation request definite edge position, then the edge position may come up the estimate in the sub-picture element resolution, the edge position also may estimate.In the edge examination algorithm, the first three steps use extremely universally.This is because under the majority situations, needs the edge detector to point out merely the edge appears in image some picture element neighbor, but is not unnecessary to point out the edge the exact location or the direction. The edge examines the error usually is refers to the edge to classify the error by mistake, namely distinguished the vacation edge the edge retains, but distinguished the real edge the vacation edge removes.The edge error of estimation is describes the edge position and the lateral error with the probability statistical model.We examine the edge 第5页 青岛大学毕业设计英文资料翻译 the error and the edge error of estimation differentiate, is because their computational method is completely different, its error model completely is also different. The edge examination is examines the image partial remarkable change the most fundamental operation. In the unidimensional situation, the step edge concerns with the image first derivative partial peak value. The gradient is the function change one kind of measure, but an image may regard as is the image intensity continuous function sampling point array. Therefore, is similar with the unidimensional situation, the image grey level remarkable change available gradient discrete approximation function examines. The gradient is first derivative two-dimensional equivalent -like, defines for the vector as a result of each kind of reason, the image always receives the stochastic noise the disturbance, may say the noise is ubiquitous.Because the classical edge examination method has introduced each form differentiate, thus causes inevitably to the noise extremely sensitive, carries out the edge examination result is frequently examines the noise regard peripheral point, but but the genuine edge also as a result of receives the noise jamming not to examine.Thus regarding has the noise image, one good edge examination method should have the good noise abatement ability, simultaneously also has the complete edge maintenance characteristic。 2-11 looks by Figure place, does not have in the noise situation in the image, the Prewitt operator, the Robert operator, the Sobel operator as well as the differential gradient operator, all can the quite accurate examination edge.But, after joins the white gaussian noise, the Robert operator receives the influence is smallest, next is the Prewitt operator, receives affects in a big way is the Sobel operator, but regarding the differential gradient operator, then is the image overall contrast gradient has obvious depression.2-17 may see by Figure, in does not have in the noise situation, the Canny operator, the LOG operator and the Laplace operator all may obtain the quite good examination effect, but, the LOG operator always can produce the false edge, this and its zero crossing examination method concerns.After adds on the noise, traditional examination operator (Laplace operator) the examination quality dropped obviously, but the LOG operator has produced more false edges under the noise condition.But the Canny operator About the noise to the edge examination result is continuously extremely satisfying. algorithm influence, we in the next section, take the LOG algorithm as the example, makes the quite exhaustive analysis and the elaboration. This chapter first from the tradition based on image gradation first derivative edge obtaining, introduced the classics edge examination operator as well as afterwards developed the linear filtering edge examination method, compared with they respective characteristic, as well as in each kind of noise situation examination ability, and has given the simulation result, finally discovered, the Canny operator and withdrew the edge aspect in the noise elimination to have the quite good effect.In this chapter final, but also take the LOG operator as the example, has analyzed the image noise and the edge shape to its examination result analysis, and has given the simulation result, explained under the traditional fixed criterion examination method, the image edge symmetry as well as the criterion choice to examines the result to have the very tremendous influence.Finally, we according to Amlan the Kundu recommendation 20 kind of template shapes, examines in the LOG algorithm in the edge image, withdraws the positive and negative boundary the central position, takes our final two value single picture element boundary output. Third chapter based on wavelet thought image edge examination Although the edge detection had the differential gradient operator, the Laplace operator, the Sobel operator, the LOG operator as well as the Canny operator and so on 第6页 青岛大学毕业设计英文资料翻译 many methods, but these algorithms do not have the automatic focal variation thought.But in fact, as a result of reasons and so on physics and illumination, in each image edge usually produces in the different criterion scope, forms the different type the edge (for example edges and so on step, roof), these informations are unknown.Moreover, in the image always has the noise, therefore, according to the image characteristic, can auto-adapted Easy to imagine, is not examine the image the edge is correctly extremely difficult. impossible to examine all edges with the sole criterion edge examination operator, simultaneously, for avoids affecting the edge examination during filtration noise the accuracy, examines the edge with the multi-criterion method more and more to bring to people's attention.Because the wavelet transformation has the good time frequency localization characteristic as well as multi-criterion analysis ability [10,11,12], has “the focal variation” in the different criterion the function, suits in the examination sudden change signal, is examines this kind of signal the powerful tool, therefore obtained the widespread application.This chapter on take the wavelet transformation as the foundation, after the use wavelet filter the image characteristic, in sampling and in the non-sampling foundation, the examination image edge, and gave some improvement comment.Based on the non-sampling wavelet transformation edge detection, the spatial information which the maintenance optimizes, has obtained the quite good effect, but, when we merely are interested to the image approximate outline time, the sampling wavelet transform pair we are perhaps more practical. The sampling wavelet transformation is the Mallat tower system algorithm which the face explains things in front of this chapter mentioned, it can obtain the primitive image in different criterion detail.When the decomposition criterion increases, not only has filtered the noise, but also withdrew the image approximate outline.We may use under the great criterion the edge image to instruct the small criterion lower limb the extraction.Thus, has removed the small detail which the very sensitive majority of noises and we is not cared about very much under the small criterion, is only the image main outline which retains us to care about.Its flow chart as shown in Figure 3-7.Must pay attention in here, the transformation criterion do not have too to be big, otherwise made the instruction the edge too to be short, easy to create the edge information to lose too many, the outline was stiff, was not gentle. And after the decomposition criterion excessively are many, a shifting can cause slightly in the great criterion in the small criterion the very big mistake.Among them, this algorithm difficulty provides the instruction in the great criterion edge for the small criterion the part.Because the great criterion edge will be sampling later the obtained data, therefore above it in a point correspondence and small criterion four spots.If in small criterion four spot intensity enough big, we on determine it as the boundary point.This algorithm simulation result as shown in Figure 3-8.In here, our have not used the wavelet inverse transformation which in the usual method uses, but, has still obtained the quite good result. ?3.5 this chapter section This chapter in after introduction wavelet edge examination principle, application, introduced uses in scoring the signal irregularity Lip index, and explained under certain condition, also may estimate function f (t) with the wavelet transformation in a t0 Lip index.Thus, related the wavelet transformation and the signal Lip index.When carries on the edge examination using the wavelet, applies non-sampling the wavelet transformation, and has made some improvements to the traditional method, unifies the auto-adapted filter method with the signal irregularity examination, in the noise quite big situation, has obtained the quite satisfying denoising effect.In this chapter final, but also discussed 第7页 青岛大学毕业设计英文资料翻译 in the sampling wavelet transformation foundation, carries on the image the main edge extraction method.This method is takes with the great criterion under edge image instructs under the small criterion the edge image to carry on the choices, the simulation result indicated this method is effective. Fourth chapter based on phase information image edge examination algorithm We knew that, the image characteristic examination is imagery processing, pattern recognition, based on content domain and so on image retrieval key technologies.How examines as well as the effective description image characteristic, since long ago receives the multitudinous disciplines continuously the attention.The overwhelming majority image characteristic examination algorithm all is carries on based on the image brightness gradient.Notes the phase information the importance and the stability, causes to carry on the imagery processing into new research topic using the phase information.This chapter, we will launch the discussion on this quite novel topic, will infer its rationale, and explained by its simulation result based on the phase characteristic extraction algorithm superiority. ?4.1 signal phase information We knew that, classical, the simple edge examination method is inspects in its some neighborhood to primitive image each picture element the gradation change, uses nearby the edge a step or two step directional derivative change rule examines the edge, for example the differential gradient operator, the Sobel operator, the Laplace operator, the Canny operator and so on, their common characteristic is: To the primitive image according to the picture element some neighborhood structure edge, examines the image partial characteristic discontinuity.In the two-dimensional picture edge examination, the Laplace operator is simple and the effective method. Because the image receives the stochastic noise frequently the disturbance, when carries on the edge examination with these operators, although has the computation load small merit, but by the differentiate inherent characteristic, they is extremely sensitive to the noise jamming, like this they when carry on the edge examination can regard as frequently the noise the edge to examine, but but the genuine edge as a result of has also not been examined the noise jamming.Therefore, obtains the edge often has flaws and so on gap, burr. Considered the phase information the importance and the stability, cause using the phase information to carry on the imagery processing into new research topic, here proposed the phase uniformity, is precisely based on this realization.Phase uniformity (PC:Phase Congruency) the basic concept is the image Fourier component phase most consistent spot achievement characteristic point.When for example, the square-wave launches for the Fourier's series, all Fourier component all is a sine wave, as shown in Figure 4-1.In a step synchronism, the phase is 90o or 270o, all is changing in the square-wave other single phase value, causes the phase coincidence degree to reduce.Similar, the phase coincidence degree is biggest in the triangular wave apex, as shown in Figure 4-2, namely 0o or 180o are biggest.The use phase uniform important characteristic is does not need to the profile to carry on any supposition, only is simply phase coincidence seeks the characteristic point in the Fourier transformation territory according to. Each kind of characteristic types and so on step, line, roof all may cause the phase uniformity high spot appearance, thus obtains the examination result, even can examine the Mach belt phenomenon which produces as a result of the human vision sensation characteristic. ?4.2 signal phase characteristic 第8页 青岛大学毕业设计英文资料翻译 The phase uniformity is the performance extremely outstanding image characteristic examination operator, has the very strong versatility, also conforms to the human vision sensation characteristic.Oppenheim and Lim[22] have given the phase information important classical demonstration.They carry on separately two charts the Fourier transformation, takes chart the phase information, again takes other chart the scope information, combines a new image, and makes the inverse Fourier transform.We may discover, on new synoptic map, although the image stripe information quite is disorderly, but, may clear seeing provide the phase information the image outline.This demonstration result as shown in Figure 4-3.Since the image phase information had been proven it has the so vital role in the entire image, the nature we can to give the quite key attention based on the image phase information image characteristic extraction algorithm. In Morrone and Owens[23] conducts after the research to the Mach belt phenomenon, confirmed the phase information in image characteristic examination many superiority, and causes this method more widespread application in the machine vision this chapter first two image Fourier transformation, exchanges the two separately the scope and the phase information, has obtained two synoptic maps.Explained the phase information by this experiment in the image importance.Then the introduction phase uniform definition, its approximate expression - partial energy as well as based on the wavelet phase uniformity computational method, has given the computation step simulation result.Meets down, after has compared the Gabor wavelet, explained log the Gabor wavelet merit, and take log the Gabor wavelet as a foundation, has designed the 2D filter, has analyzed its performance, and through has proven our analysis to each kind of image examination result.Finally has given the phase uniform algorithm to the Mach belt phenomenon processing result, explained this algorithm conforms to humanity's visual system performance. Full text summary Without a doubt, in the information highly developed modern society, seeps along with the information technology to the production and life each domain.Pattern recognition technology application is more widespread.But the image edge detection is precisely the pattern recognition, application and so on machine vision foundations and the premise.This paper has carried on the review on the present quite mature classics edge examination algorithm, and applied many to at present has made the introduction based on the wavelet characteristic examination algorithm, finally, but also introduced recently developed by the image phase information extraction edge technology.Generally speaking, this article very clear divides into three parts, between them both relative independence, and layer upon layer progressives. The first part has given a total explanation to the present image edge examination algorithm, elaborated the paper selected topic theory significance.The second part first from the tradition based on image gradation first derivative edge obtaining, introduced the classics edge examination operator as well as afterwards developed linear filtering edge examination method, compared with they respective characteristic, as well as in each kind of noise situation examination ability.In this part final, but also take the LOG operator as the example, has analyzed the image noise and the edge shape to its examination result analysis, explained under the traditional fixed criterion examination method, the image edge symmetry as well as the criterion choice to examines the result to have the very tremendous influence.Finally, we according to Amlan the Kundu recommendation 20 kind of template shapes, examines in the LOG algorithm in the edge image, withdraws the positive and negative boundary the central position, the achievement final two value single picture element boundary output. 第9页 青岛大学毕业设计英文资料翻译 After third part of introduction wavelet edge examination principle, application, introduced uses in scoring the signal irregularity Lip index, and explained under certain condition, also may estimate function f (t) with the wavelet transformation in a t0 Lip index.When carries on the edge examination using the wavelet, applies non-sampling the wavelet transformation, and has made some improvements to the traditional method, unifies the auto-adapted filter method with the signal irregularity examination, in the noise quite big situation, has obtained the quite satisfying denoising effect.In this chapter final, but also discussed in the sampling wavelet transformation foundation, carries on the image the main edge extraction method.This method is takes with the great criterion under edge image instructs under the small criterion the edge image to carry on the choices, the simulation result indicated this method is effective. Fourth introduction phase uniform definition, its approximate expression - partial energy as well as based on wavelet phase uniformity computational method.After has compared the Gabor wavelet, explained the logGabor wavelet merit, and take log the Gabor wavelet as a foundation, has designed the 2D filter, has analyzed its performance, and through has proven our analysis to each kind of image examination result.Finally has given the phase uniform algorithm to the Mach belt phenomenon processing result, explained is the algorithm conforms to humanity's vision based on the phase image characteristic examination algorithm system performance. The above is this paper roughly the structure and the primary coverage, as well as to a certain algorithm small improvement.Through master the stage study and the topic research, caused me to open up the aspect of knowledge, has raised certain theory and engineering research ability, and could understand thoroughly in modern science and technology multi-disciplinary, multi-domain mutual seepage, unified mutually relations, also caused me basically to grasp objective, comprehensive and the scientific research question method, for will continue to carry on the next step research work from now on to build the solid foundation.Finally, the author comments in this depth deep thanks for this article reads pays the time and the pain fellow experts! 第10页 青岛大学毕业设计英文资料翻译 图像边缘检测算法 摘 要 数字图像处理作为一门相对比较年轻的学科，伴随着计算机技术的飞速发展，日益得到广泛的 应用。边缘作为图像的一种基本特征，在图像识别,图像分割,图像增强以及图像压缩等的领域中有 较为广泛的应用。图像边缘提取的手段多种多样，其中基于亮度的算法，是研究时间最久，理论发展最成熟的方法，它主要是通过一些差分算子，由图像的亮度计算其梯度的变化，从而检测出边缘，主要有Robert, Laplacian, Sobel, Canny, LOG 等算子。 首先从总体上介绍了数字图像处理及边缘提取的概况，列举了几种目前常用的边缘提取技术和算法,并选取其中两种使用Visual C++语言编程实现，通过对两种算法所提取图像结果的比较，研究探讨它们的优缺点。 前 言 在实际图像处理问 题 快递公司问题件快递公司问题件货款处理关于圆的周长面积重点题型关于解方程组的题及答案关于南海问题 中，图像的边缘作为图像的一种基本特征，经常被应用到较高层次的图像应用中去。它在图像识别，图像分割，图像增强以及图像压缩等的领域中有较为广泛的应用，也是它们的基础。图像边缘检测的手段多种多样。但是，其大的框架不外乎两种，即传统的基于图像亮度特征的算法和基于小波的多尺度边缘检测算法。对于基于亮度的算法，是研究时间最久，理论发展最成熟的方法，它主要是通过一些差分算子，由图像的亮度计算其梯度的变化，从而检测出边缘，主要有Robert,Laplace,Sobel,Canny,LOG 等算子。这些算法现在已经发展的比较成熟了. 再有一类就是随这小波理论的发展和成熟而兴起的基于小波变换的多尺度的图像边缘检测算法。它利用小波变换，检测出图像的在行方向上和列方向上的跃变边缘，并在一定的规则下形成图像的边缘;还有一种方法就是利用小波的多尺度变换以及在小波变换下信号和噪声的 Lip 指数的区别来提取对我们有用的边缘信息，并除去噪声的干扰。不过近几年来，发展出了一种新颖的基于图像像素相位的边缘检测算法。它不需要对图像进行任何先验的假设，只是在傅立叶变换域内简单的按相位一致来寻找特征点，鲁棒性比较强。 本文在第一章对图像的边缘进行了一个一般的概述，接着，在第二章中先介绍了比较经典的检测算子，并以 LOG 算法为例子，介绍了定位误差产生的原因，讨论了一种比较精确的边缘定位方法。在第三章中，在介绍小波理论的基础上，分别介绍了抽样小波变换和非抽样小波变换在边缘检测中的应用。其中非抽样的小波变换侧重于保持图像优化的空间信息和利用 Lip 指数去除噪声，而抽样的小波变换则方便我们提取图像的主要轮廓。在本文的最后一章，讨论了基于相位的边缘提取算法。利用 log Gabor 小波，构造了 2D滤波器，成功的检测出了多种图像的边缘，甚至还检测出了马赫带现象，说明这是一种优秀的检测方法，并且符合人类的视觉系统特性。相信本文的工作对图像处理中的边缘检测方法研究以及应用有一定的参考价值。 第一章 绪论 ?1.1 图像边缘检测概论 图像边缘是图像最基本的特征之一，往往携带着一幅图像的大部分信息。而边缘存在于图像的不规则结构和不平稳现象中，也即存在于信号的突变点处，这些点给出了图像轮廓的位置，这些轮廓常常是我们在图像处理时所需要的非常重要的一些特征条件，这就需要我们对一幅图像检测并提取出它的边缘。而边缘检测算法则是图像处理问题中经典技术难题之一，它的解决对于我们进行高层次的特征描述、识别和理解等有着重大的影响;又由于边缘检测在许多方面都有着非常重要的使用价值，所以人们一直在致力于研究和解决如何构造出具有良好性质及好的效果的边缘检测算子的问题。在通常情况下，我们可以将信号中的奇异点和突变点认为是图像中的边缘点，其附近灰度的变化情况可从它相邻像素灰度分布的梯度来反映。 第11页 青岛大学毕业设计英文资料翻译 根据这一特点，我们提出了多种边缘检测算子:如 Robert算子、Sobel 算子、Prewitt 算子、Laplace 算子等。这些方法多是以待处理像素为中心的邻域作为进行灰度分析的基础，实现对图像边缘的提取并已经取得了较好的处理效果。但这类方法同时也存在有边缘像素宽、噪声干扰较严重等缺点，即使采用一些辅助的方法加以去噪，也相应的会带来边缘模糊等难以克服的缺陷。随着小波分析的出现，其良好的时频局部特性被广泛的应用在图像处理和模式识别领域中，成为信号处理中常用的手段和有力的工具。通过小波分析，可以将交织在一起的各种混合信号分解成不同频率的块信号，而通过小波变换进行边缘检测，可以充分利用其多尺度和多分辨率的性质，真实有效的 表 关于同志近三年现实表现材料材料类招标技术评分表图表与交易pdf视力表打印pdf用图表说话 pdf 达图像的边缘特征。当小波变换的尺度减小时，对图像的细节更加敏感;而当尺度增大时，图像的细节将被滤掉，检测的边缘只是粗轮廓。该特性在模式识别中非常有用，我们可以将此粗轮廓称为图像的主要边缘。如果能将一个图像的主要边缘清晰完整的提取出来，这将对目标分割、识别等后续处理带来极大的便利。总的说来，以上方法都是基于图像的亮度信息来作的工作。 在众多科研工作者的努力下，取得了很好的效果。但是，由于图像边缘受到光照等物理条件的影响比较大，往往使得以上诸多基于亮度的边缘提取方法有着一个共同的缺点，那就是边缘不连续，不封闭。考虑到相位信息在图像中的重要性以及其稳定的特点，使得利用相位信息进行图像处理成为新的研究课题。 在本文中即将介绍一种基于相位的图像特征检测方法——相位一致性方法。它并不是利用图像的亮度信息，而是其相位特点，即假设图像的傅立叶分量相位最一致的点作为特征点。它不但能检测到阶跃特征、线特征等亮度特征，而且能够检测到由于人类视觉感知特性而产生的的马赫带现象。由于相位一致性不需要对图像的特征类型进行任何假设，所以它具有很强的通用性。 ?1.2 图像边缘的定义 图像的大部分主要信息都存在于图像的边缘中，主要表现为图像局部特征的不连续性，是图像中灰度变化比较剧烈的地方，也即我们通常所说的信号发生奇异变化的地方。奇异信号沿边缘走向的灰度变化剧烈，通常我们将边缘划分为阶跃状和屋顶状两种类型(如图 1-1所示)。阶跃边缘中两边的灰度值有明显的变化;而屋顶状边缘位于灰度增加与减少的交界处。在数学上可利用灰度的导数来刻画边缘点的变化，对阶跃边缘、屋顶状边缘分别求其一阶、二阶导数。 对一个边缘来说，有可能同时具有阶跃和线条边缘特性(例如在一个表面上，由一个平面变化到法线方向不同的另一个平面就会产生阶跃边缘;如果这一表面具有镜面反射特性且两平面形成的棱角比较圆滑，则当棱角圆滑表面的法线经过镜面反射角时，由于镜面反射分量，在棱角圆滑表面上会产生明亮光条，这样的边缘看起来象在阶跃边缘上叠加了一个线条边缘(由于边缘可能与场景 第12页 青岛大学毕业设计英文资料翻译 中物体的重要特征对应，所以它是很重要的图像特征。比如，一个物体的轮廓通常产生阶跃边缘， 因为物体的图像强度不同于背景的图像强度。 ?1.3 论文选题的理论意义 论文选题来源于在图像工程中 占有重要的地位和作用的实际应用课题。所谓图像工程学科是 指将数学、光学等基础学科的原理，结合在图像应用中积累的技术经验而发展起来的学科。图像工程的 内容 财务内部控制制度的内容财务内部控制制度的内容人员招聘与配置的内容项目成本控制的内容消防安全演练内容 非常丰富，根据抽象程度和研究方法等的不同分为三个层次:图像处理，图像分析和图像 理解。如图 1-2 所示，在图中，图像分割处于图像分析与图像处理之间，其含义是，图像分割是从图像处理进到图像分析的关键步骤，也是进一步理解图像的基础。 图像分割对特征有重要影响。图像分割及基于分割的目标表达、特征提取和参数测量等将原始图像转化为更抽象更紧凑的形式，使得更高层的图像分析和理解成为可能。而边缘检测是图像分割的核心内容，所以边缘检测在图像工程中占有重要的地位和作用。因此边缘检测的研究一直是图像技术研究中热点和焦点，而且人们对其的关注和投入不断提高。 ?1.4 本文主要工作 ?1.4.1 算法内容 介绍和分析了经典的图像边缘检测算法，总结了各个算法的优缺点，给出了图像边缘的检测结果，并着重以 LOG 算法为例，从噪声和边缘形态的观点出发分析了边缘定位误差产生的原因;介绍了一种 LOG 算法中比较精确的确定边缘的方法。 ?1.4.2 小波理论 研究了小波的基本理论，总结了信号以及噪声的 Lip 指数的性质，并在基于非抽样小波变换的图像特征提取算法的基础上，结合自适应的去噪方法，对该方法作出了一定的改进，取得了比较满意的效果，去噪能力有了比较大的提高;接着介绍了一种基于抽样小波检测图像主要边缘的方法。 ?1.4.2 新颖算法 系统研究了一种比较新颖的基于相位的图像特征提取算法——相位一致性算法，并给出了其简便算法。给出了在一维情况下的算法仿真步骤，分析了扩展到二维方法，并由边缘检测结果说明了相位一致性算法更符合人类视觉的特征。 ?1.5 本文内容安排 在第一章，作者对图像边缘检测作了一个概要的说明，并说明了进行图像边缘检测的重要意义。第二章中，系统介绍了比较经典的图像边缘检测算子及其具体的实现原理，由边缘检测的结果分析了各个算法存在的不足。最后，从噪声影响和边缘的形态入手，以 LOG 算法为例，分析了虚假边缘以及边缘移位产生的原因。最后介绍了一种 LOG 算法中比较精确的确定边缘的方法。 第三章中，作者系统介绍了目前比较流行的小波理论，并着重介绍了多尺度的概念和信号的 Lip 指数，并由噪声和信号 Lip 指数的特点，在非抽样小波变换的基础上对图像的边缘进行提取。为了加强边缘图像的抗噪能力，还对此算法进行了一定的改进，将自适应的去噪方法用到了边缘提取中，取得了令人满意的效果。最后还介绍一种基于抽样小波检测图像主要边缘的方法。在本文的第四章，介绍了一种比较新颖的基于相位的图像特征提取算法——相位一致性算法。从一维的算法 第13页 青岛大学毕业设计英文资料翻译 介绍入手，给出了一维信号下的仿真结果，并逐步扩展到二维。通过仿真结果说明了此算法的鲁棒 性比较强，而且符合人类的视觉系统特性。 第二章 经典图像边缘检测算法 本章首先在 2.1 节简单介绍了经典一阶边缘检测的基本方法。然后的 2.2 节和 2.3 节分别 具体阐述了经典微分算子和线性滤波算子的实现方法，并在 2.4 节给出了各种算法的结果比较。 2.5 节，比较具体的分析了噪声和边缘形态对边缘定位精度的影响以及伪边缘产生的原因，并在 给出了一维情况下的仿真结果，得出了结论。在 2.6 节中，利用 LOG 算法检出的图像的正负边缘，比较精确的定位出了图像的真实边缘，最后输出的是二值单像素图像。 ?2.1 经典边缘检测的基本方法 我们知道，边缘检测的实质是采用某种算法来提取出图像中对象与背景间的交界线。我们将边缘定义为图像中灰度发生急剧变化的区域边界。图像灰度的变化情况可以用图像灰度分布的梯度来反映，因此我们可以用局部图像微分技术来获得边缘检测算子。 边缘检测算法有如下四个步骤(其过程如图 2-1 所示): 滤波:边缘检测算法主要是基于图像强度的一阶和二阶导数，但导数的计算对噪声很敏感，因此必须使用滤波器来改善与噪声有关的边缘检测器的性能。需要指出，大多数滤波器在降低噪声的同时也导致了边缘强度的损失，因此，增强边缘和降低噪声之间需要折衷。 增强:增强边缘的基础是确定图像各点邻域强度的变化值。增强算法可以将邻域(或局部)强度值有显著变化的点突显出来。边缘增强一般是通过计算梯度幅值来完成的。 检测:在图像中有许多点的梯度幅值比较大，而这些点在特定的应用领域中并不都是边缘，所以应该用某种方法来确定哪些点是边缘点。最简单的边缘检测判据是梯度幅值阈值判据。 定位:如果某一应用场合要求确定边缘位置，则边缘的位置可在子像素分辨率上来估计，边缘的方位也可以被估计出来。在边缘检测算法中，前三个步骤用得十分普遍。这是因为大多数场合下，仅仅需要边缘检测器指出边缘出现在图像某一像素点的附近，而没有必要指出边缘的精确位置或方向。 边缘检测误差通常是指边缘误分类误差，即把假边缘判别成边缘而保留，而把真边缘判别成假边缘而去掉。边缘估计误差是用概率统计模型来描述边缘的位置和方向误差的。我们将边缘检测误差和边缘估计误差区分开，是因为它们的计算方法完全不同，其误差模型也完全不同。 边缘检测是检测图像局部显著变化的最基本运算(在一维情况下，阶跃边缘同图像的一阶导数局部峰值有关(梯度是函数变化的一种度量，而一幅图像可以看作是图像强度连续函数的取样点阵列(因此，同一维情况类似，图像灰度值的显著变化可用梯度的离散逼近函数来检测(梯度是一阶导数的二维等效式，定义为向量由于各种原因，图像总是受到随机噪声的干扰，可以说噪声无处不在。经典的边缘检测方法由于引入了各种形式的微分运算，从而必然引起对噪声的极度敏感，执行边缘检测的结果常常是把噪声当作边缘点检测出来，而真正的边缘也由于受到噪声干扰而没有检测 第14页 青岛大学毕业设计英文资料翻译 出来。因而对于有噪声图像来说，一种好的边缘检测方法应该具有良好的噪声抑制能力，同时又有 完备的边缘保持特性。 由图 2-11 看处，在图像没有噪声的情况下，Prewitt 算子、Robert算子、Sobel 算子以及 微分梯度算子，都能够比较准确的检测出边缘。但是，当加入高斯白噪声后，Robert 算子受到的影响最小，其次为Prewitt 算子，受到影响最大的是 Sobel 算子，而对于微分梯度算子来说，则 是图像的整体对比度有明显的降低。由图 2-17 可以看出，在没有噪声的情况下，Canny 算子、LOG算子和 Laplace 算子都可以得到比较好的检测效果，但是，LOG 算子总是会产生伪边缘，这和其过零点的检测方法有关。在加上噪声后，传统检测算子(Laplace 算子)的检测质量明显下降了，而 LOG 算子在噪声条件下则产生了更多的伪边缘。但是 Canny 算子的检测结果一直非常令人满意。关于噪声对边缘算法的影响，我们将在下一小节中，以 LOG 算法为例，作出比较详尽的分析和阐述。 本章首先从传统的基于图像灰度的一阶导数的边缘入手，介绍了经典的边缘检测算子以及后来发展起来的线性滤波边缘检测方法，比较了它们各自的特点，以及在各种噪声情况下的检测能力，并给出了仿真的结果，最后发现，Canny 算子在抑制噪声和提取边缘方面有比较好的效果。在本章的最后，还以 LOG 算子为例，分析了图像的噪声和边缘的形态对其检测结果的分析，并给出了仿真结果，说明在传统的固定尺度的检测手段下，图像边缘的对称性以及尺度的选择对检测结果有很大的影响。最后，我们根据 Amlan Kundu 推荐的二十种模板形态，在 LOG 算法检测出的边缘图像中，提取出正负边界的中心位置，作为我们的最后的二值单像素边界输出。 第三章 基于小波思想的图像边缘检测 虽然边缘提取已有微分梯度算子、Laplace 算子、Sobel 算子、LOG 算子以及 Canny 算子等诸多方法，但是这些算法都没有自动变焦的思想。而事实上，由于物理和光照等原因，每幅图像中的边缘通常产生在不同的尺度范围内，形成不同类型的边缘(如阶跃、屋顶等边缘)，这些信息是未知的。另外，图像中总是存在噪声的，因此，根据图像特性，能够自适应的正确检测出图像的边缘是非常困难的。容易想象，用单一尺度的边缘检测算子是不可能检测出所有的边缘的，同时，为避免在滤除噪声的同时影响边缘检测的正确性，用多尺度的方法检测边缘越来越引起人们的重视。由于小波变换具有良好的时频局部化特性以及多尺度分析能力[10,11,12]，在不同尺度上具有“变焦”的功能，适合于检测突变信号，是检测这类信号的有力工具，所以得到了广泛的应用。本章就将以小波变换为基础，利用小波滤波后图像的特性，在抽样和非抽样的基础上，检测图像边缘，并提出了一些改进意见。基于非抽样小波变换的边缘提取，保持优化的空间信息，取得了比较好的效果，但是，当我们仅仅对图像的大致轮廓感兴趣的时候，抽样的小波变换对我们来说也许更实用。抽样的小波变换就是在本章前面部分说提到的 Mallat 塔式算法，它能够得到原始图像在不同尺度上的细节。当分解尺度增大时，不仅滤除了噪声，还提取出了图像的大致轮廓。我们可以用大尺度下的边缘图像指导小尺度下边缘的提取。这样，去除了在小尺度下很敏感的大部分噪声和我们所不是很关心的微小细节，只是保留我们关心的图像主要轮廓。其流程图如图 3-7 所示。在这里要注意的是，变换的尺度不要太大，否则作指导的边缘太少了，容易造成边缘信息丢失太多，轮廓生硬，不柔和。并且分解尺度过多后，在大尺度上的稍微一点移位都会导致在小尺度上的很大差错。其中，本算法的难点在大尺度边缘为小尺度提供指导的部分。由于大尺度边缘是抽样以后所得的数据，所以它上面的一个点对应与小尺度上的四个点。如果小尺度上的四个点的强度足够的大，我们就将其判定为边界点。本算法仿真结果如图 3-8 所示。在这里，我们没有使用通常方法中用到的小波反变换，但是，仍然取得了比较好的结果。 ?3.5 本章小节 第15页 青岛大学毕业设计英文资料翻译 本章在介绍小波边缘检测的原理、应用以后，介绍了用于刻划信号奇异性的 Lip 指数，并说 明了在一定的条件下，用小波变换也可以估计函数 f (t)在一点 t0 的 Lip 指数。这样，就把小 波变换和信号的 Lip指数联系起来了。在应用小波进行边缘检测时，应用非抽样的小波变换，并 且对传统的方法作出了一些改进，用信号奇异性的检测结合自适应的滤波方法，在噪声比较大的情 况下，取得了比较令人满意的去噪效果。在本章的最后，还讨论了在抽样小波变换的基础上，进行 图像的主要边缘的提取的方法。此方法是用大尺度下的边缘图像取指导小尺度下的边缘图像进行取 舍，仿真结果表明该方法是有效的。 第四章 基于相位信息的图像边缘检测算法 我们知道，图像特征检测是图像处理、模式识别、基于内容的图像检索等领域的关键技术。如何检测以及有效的描述图像的特征，长期以来一直受到众多学科的关注。绝大多数的图像特征检测算法都是基于图像亮度梯度进行的。注意到相位信息的重要性和稳定性，使得利用相位信息进行图像处理成为新的研究课题。本章，我们将就这一比较新颖的课题展开讨论，推导其理论基础，并由其仿真结果说明了基于相位的特征提取算法的优越性。 ?4.1 信号的相位信息 我们知道，经典的，最简单的边缘检测方法是对原始图像的每个像素考察它的某个邻域内灰度的变化，利用边缘附近的一阶或二阶方向导数变化规律检测边缘，例如微分梯度算子，Sobel 算子，Laplace 算子，Canny 算子等，它们的共同特性是:对原始图像按像素的某邻域构造边缘，检测出图像局部特性的不连续性。在二维图像边缘检测中，Laplace 算子是最简单而有效的方法。由于图像常常受到随机噪声的干扰，当用这些算子进行边缘检测时，虽然具有计算量小的优点，但由微分运算固有的特性，它们对噪声干扰十分敏感，这样它们在进行边缘检测时常常会把噪声当成边缘检测出来，而真正的的边缘也由于受噪声干扰而没有被检测出来。因此，得到的边缘往往有缺口、毛刺等缺陷。 考虑到相位信息的重要性和稳定性，使得利用相位信息进行图像处理成为新的研究课题，这里提出的相位一致性，正是基于此实现的。相位一致性(PC:Phase Congruency)的基本概念是将图像傅立叶分量相位最一致的点作为特征点。例如，方波展开为傅立叶级数时，所有的傅立叶分量都是正弦波，如图 4-1 所示。在阶跃点同相，相位为 90o或 270o，在方波的其它点的单个相位值都在变化，使得相位一致的 程度降低。类似的，相位一致程度在三角波的顶点最大，如图 4-2 所示，即 0o或 180o最大。使用相位一致性的重要特点是无需对波形进行任何假设，只是在傅立叶变换域里简单按相位一致来寻找特征点。各种阶跃、线、屋顶等特征类型都可以使得相位一致性高的点出现，从而得到检测结果，甚至能够检测到由于人类视觉感知特性而产生的马赫带现象。 ?4.2 信号的相位特性 相位一致性是性能非常优秀的图像特征检测算子，具有很强的通用性，且符合人类视觉感知特性。Oppenheim 和 Lim[22]给出了相位信息重要性的经典示例。他们分别将两幅图进行傅立叶变换，取其中一幅图的相位信息，再取另外一幅图的幅度信息，重新组合成一幅新的图像，并作傅立叶反变换。我们可以发现，在新的合成图上，虽然图像的条纹信息比较凌乱，但是，还是可以清晰的看见提供相位信息的图像的轮廓。这个示例结果如图 4-3 所示。既然图像的相位信息被证明了其在整个图像中有如此重要的作用，自然我们就会对基于图像相位信息的图像特征提取算法给予比较重点的关注。 在 Morrone和 Owens[23]对马赫带现象进行研究以后，证实了相位信息在图像特征检测中的诸多优越性，并使得这一方法更加广泛的应用于机器视觉中本章首先将两幅图像傅立叶变换，分别交换二者的幅度和相位信息，得出了两幅合成图。由这个实验说明了相位信息在图像中的重要性。然后介绍相位一致性的定义、其近似表示—局部能量以及基于小波的相位一致性计算方法，给出了 第16页 青岛大学毕业设计英文资料翻译 计算步骤的仿真结果。接下来，在比较了 Gabor 小波后，说明了 log Gabor 小波的优点，并以 log Gabor 小波为基础，设计了 2D 滤波器，分析了其性能，并通过对各种图像的检测结果证明了我们 的分析。最后给出了相位一致性算法对马赫带现象的处理结果，说明了本算法符合人类的视觉系统 特性。 全文总结 毫无疑问,在信息高度发达的现代社会中，随着信息技术渗透到生产和生活的各个领域。其中 的图像识别技术的应用更为广泛。而图像的边缘提取正是图像识别，机器视觉等应用的基础和前提。本篇论文就现在比较成熟经典的边缘检测算法进行了回顾，并对目前应用较多的基于小波的特征检测算法作了介绍，最后，还介绍了新近发展起来的由图像的相位信息提取边缘的技术。总的来说，本文很清晰的分为三个部分，他们之间既相对独立，又层层递进。 第一部分是对目前的图像边缘检测算法作了一个总的说明，阐述了论文选题的理论意义。第二部分首先从传统的基于图像灰度的一阶导数的边缘入手，介绍了经典的边缘检测算子以及后来发展起来的线性滤波边缘检测方法，比较了它们各自的特点，以及在各种噪声情况下的检测能力。在本部分的最后，还以 LOG 算子为例，分析了图像的噪声和边缘的形态对其检测结果的分析，说明在传统的固定尺度的检测手段下，图像边缘的对称性以及尺度的选择对检测结果有很大的影响。最后，我们根据 Amlan Kundu 推荐的二十种模板形态，在 LOG 算法检测出的边缘图像中，提取出正负边界的中心位置，作为的最后的二值单像素边界输出。 第三部分介绍小波边缘检测的原理、应用以后，介绍了用于刻划信号奇异性的 Lip 指数，并说明了在一定的条件下，用小波变换也可以估计函数 f (t)在一点 t0 的 Lip 指数。在应用小波进行边缘检测时，应用非抽样的小波变换，并且对传统的方法作出了一些改进，用信号奇异性的检测结合自适应的滤波方法，在噪声比较大的情况下，取得了比较令人满意的去噪效果。在本章的最后，还讨论了在抽样小波变换的基础上，进行图像的主要边缘的提取的方法。此方法是用大尺度下的边缘图像取指导小尺度下的边缘图像进行取舍，仿真结果表明该方法是有效的。 第四部介绍相位一致性的定义、其近似表示—局部能量以及基于小波的相位一致性计算方法。在比较了 Gabor 小波后，说明了 logGabor 小波的优点，并以 log Gabor 小波为基础，设计了 2D 滤波器，分析了其性能，并通过对各种图像的检测结果证明了我们的分析。最后给出了相位一致性算法对马赫带现象的处理结果，说明了基于相位的图像特征检测算法是算法符合人类的视觉系统特性的。 以上就是本篇论文的大体结构和主要内容，以及对某些算法的一点小小的改进。通过硕士阶段的学习和课题研究，使我拓宽了知识面，培养了一定的理论和工程研究的能力，并能较深入地了解到现代科技中的多学科、多领域的相互渗透，相互结合的关系，也使我基本掌握了客观、全面和科学研究问题的方法，为今后继续进行下一步研究工作打下了坚实的基础。最后，作者在此深深感谢为本文评阅付出时间和辛劳的各位专家学者～ 第17页