#### DMCA

## A Survey on Edge Detection Methods (1744)

Citations: | 6 - 0 self |

### Citations

4675 | A Computational Approach to Edge Detection.”
- Canny
- 1986
(Show Context)
Citation Context ...e of the image intensity function. This is theoretically the same as using the maxima in the first directional derivatives, and in one dimension is the same as the LOG filter. Later, in 80s, Canny [6]=-=[7]-=- proposed a method that was widely considered to be the standard edge detection algorithm in the industry, and still it outperforms many of recent algorithms. In regard to regularization explained in ... |

1887 | Scale-space and edge detection using anisotropic diffusion,
- Perona, Malik
- 1990
(Show Context)
Citation Context ...owever, in the case of linear heat equation as diffusion eradicates noise, it also blurs the edges isotropically (i.e. invariant with respect to direction). To overcome this problem, Perona and Malik =-=[37]-=- proposed a scale space representation of an image based on anisotropic diffusion. In the mathematical context, this calls for nonlinear partial differential equations rather than the linear heat equa... |

672 |
Scale-space filtering,
- Witkin
- 1983
(Show Context)
Citation Context ...ne the number of filters to use. In addition, by choosing such a large size for the smallest filter, Schunck‘s technique loses a lot of important details which may exist at smaller scales [5]. Witkin =-=[29]-=- studied the property of zero-crossings across scales for 1D signal. He marked the zero-crossings of second derivative of a signal smoothed by Gaussian function in a range of scale, and then presented... |

374 |
Digital Image Processing.
- Pratt
- 1978
(Show Context)
Citation Context ...�� � � 0 (11) r � �� g( c �1, r) �� �� �1�� ��1� H c � ��1 0 �1� � , H � r � � 0 (12) � �� �1�� Several other first order derivative approximations along two perpendicular axes have been proposed [3]-=-=[4]-=-, and some of the most known of them are as follow: Roberts: Prewitt: Sobel: � 0 �1� ��1 0 � H 1 � � � , H 2 � ��1 0 � � (13) � � 0 �1� ��1 0 �1� 1 H � � c � � �1 0 �1 , 3 � �� �1 0 �1�� ��1 0 �1� 1 H... |

200 | Digital Step Edges from Zero Crossing of the Second Directional Derivatives”,
- Haralick
- 1984
(Show Context)
Citation Context ...g algorithms based on this approach will remain heuristic in nature. Furthermore, physiological experiments have found no evidence to support zero-crossings as a model for biological vision. Haralick =-=[27]-=- proposed the use of zero-crossing of the second directional derivative of the image intensity function. This is theoretically the same as using the maxima in the first directional derivatives, and in... |

181 | Scaling theorems for zero crossings,
- Yuille, Poggio
- 1986
(Show Context)
Citation Context ...tput. It is proven that when an image is smoothed by a Gaussian filter, the existing zero-crossings (i.e. detected edges) disappear as moving from fine-to-coarse scale, but new ones are never created =-=[26]-=-. This unique property makes it possible to track zero-crossings (i.e. edges) over a range of scales, and also gives the ability to recover them at sufficiently small scales. Yuille and Poggio [26] pr... |

140 |
Edge Focusing,"
- Bergholm
- 1987
(Show Context)
Citation Context ...mallest scale to the scale at which it disappears. This work initiated the study of edge detection as a function of scale, and led to algorithms that combine edges for better edge detection. Bergholm =-=[30]-=- proposed an algorithm which uses the Gaussian filter and combines edge information moving from a coarse-to-fine scale. His method is called edge focusing, and uses a rule-based approach for detecting... |

89 | Statistical edge detection: Learning and evaluating edge cues
- Konishi, Yuille, et al.
- 2003
(Show Context)
Citation Context ...d contextual filter detects edges from the finest scale gradient images and then, the edge tracker refines the detected edges on the multiscale gradient images. 4.6 Statistical Methods Konishi et al. =-=[16]-=- formulated the edge detection as a statistical inference. This statistical edge detection is data driven, unlike standard methods for edge detection which are model based. For any set of edge detecti... |

70 | Comparison of Edge Detectors: A Methodology and Initial Study”.
- Heath, Sarkar, et al.
- 1998
(Show Context)
Citation Context ...dge detectors. Subjective and objective evaluations can be used together to evaluate edge detectors. This combination inspired by psychological methods, is based on statistical analysis. Heath et al. =-=[44]-=- proposed an evaluation method in the context of object recognition. Edge detector results were presented to subject humans who compare different edge detectors. The results were interpreted using the... |

43 |
Edge evaluation using local edge coherence
- Kithchen, Rosenfeld
- 1981
(Show Context)
Citation Context ...detection of true edges, detection of false edges, and localization error. Using this measure it is difficult to determine the type of error committed by the detector. Kitchen and Rosenfeld's measure =-=[42]-=- combines errors that arise due to an edge's thickness and lack of continuity. Venkatesh and Kitchen [43] used four error types which reflect the major difficulties encountered in edge detection: non-... |

35 | An edge detection technique using genetic algorithm based optimization, Pattern Recognit
- Bhandankar, Zhang, et al.
- 1994
(Show Context)
Citation Context ... estimation of derivatives into the both gradient and zero crossing methods to locate the edge positions. They stated a performance near to the Canny method, but faster computation. Bhandarkar et al. =-=[21]-=- presented a genetic algorithm (GA) based optimization technique for edge detection. The problem of edge detection was formulated as the choosing a minimum cost edge configuration. The edge configurat... |

30 | Finding edges and lines - Canny - 1983 |

29 |
Gaussian-Based Edge-Detection Methods—A Survey,”
- Basu
- 2002
(Show Context)
Citation Context ...away from the true edge position due to poor localization of the LOG operator, and finally there are too many zero-crossings in the small scales of a LOG filtered image, most of which is due to noise =-=[5]-=-. Relating image structures across scale is intrinsically problematic since, zero-crossings merge tangentially. Thus, one cannot rely upon zero-crossings to track edges. Edge-linking algorithms based ... |

29 | Looney," Competitive Fuzzy Edge Detection"
- Liang, Carl
- 2004
(Show Context)
Citation Context ... to the edge map and the direction map to extract the edge points. The proposed method can detect the edge successfully, while double edges, thick edges, and speckles can be avoided. Liang and Looney =-=[12]-=- introduced competitive fuzzy edge detection method. They adopted extended ellipsoidal Epanechnikov functions as a fuzzy set membership function, and a fuzzy classifier that differentiates image pixel... |

20 |
A novel edge detection method based on the maximizing objective function.
- Kang, Wanga
- 2007
(Show Context)
Citation Context ...edge detection of Sobel and Canny. (37) Figure 8 – casual template used by GAP [15] CES-506, University of Essex, U.K. Page 27Figure 9 – four direction of two sets of pixels in 3x3 mask developed in =-=[11]-=- Kang and Wang [11] developed an edge detection algorithm based on maximizing an object function. The values of the objective function corresponding to four directions determine the edge intensity and... |

20 |
Edge contours using multiple scales
- Williams, Shah
- 1990
(Show Context)
Citation Context ...e coarsest resolution that is used to determine the location of the edges. He also provides no explanation as to how to decide which scales are to be used and under what conditions. Williams and Shah =-=[32]-=- devised a scheme to find edge contours using multiple scales. They analyzed the movement of edge points smoothed with a Gaussian operator of different sizes, and used this information to determine ho... |

19 |
An adaptive Gaussian filter for noise reduction and edge detection
- Deng, Cahill
- 1993
(Show Context)
Citation Context ...oes result in reduced performance when it comes to detecting straight lines in vertical or horizontal directions. The algorithm also has the disadvantage of low-speed performance [5]. Deng and Cahill =-=[35]-=- also use an adaptive Gaussian filtering algorithm for edge detection. Their method is based on adapting the variance of the Gaussian filter to the noise characteristics and the local variance of the ... |

15 |
A geometric approach to edge detection
- Bezdek, Chandrasekhar, et al.
- 1998
(Show Context)
Citation Context ...ws that the results are not purely domain specific. They applied the same approach to the spatial grouping of edge cues and obtained analogies to non-maximal suppression and hysteresis. Bezdek et al. =-=[17]-=- described edge detection as a composition of four steps: conditioning, feature extraction, blending, and scaling. They examined the role of geometry in determining good features for edge detection an... |

14 |
Theory of edge detection Proc
- Marr, Hildreth
- 1980
(Show Context)
Citation Context ...ased edge detectors are developed based on some physiological observations and important properties of the Gaussian function that enable to perform edge analysis in the scale space. Marr and Hildreth =-=[24]-=-[25] were the pioneers that proposed an edge detector based on Gaussian filter. Their method had been a very popular one, before Canny released his detector. They originally pointed out the fact that ... |

11 |
Edge detection with Gaussian filters at mult iple scales
- Schunck
- 1987
(Show Context)
Citation Context ...n of the outputs corresponding to different scales, and adaptation to level of noise in the image. There are plenty publications in this area, we just content few sample works in this sub-section. In =-=[28]-=-, Schunck introduces an algorithm for the detection of step edges using Gaussian filters at multiple scales. The initial steps of Schunck‘s algorithm are based on Canny‘s method. The CES-506, Universi... |

10 |
The primary raster: A multi-resolution image description
- Lacroix
(Show Context)
Citation Context ... such as shadows) present a juggling effect at small scales. This is due to the splitting of a coarse edge into several finer edges, and tends to give rise to broken, discontinuous edges [5]. Lacroix =-=[31]-=- avoids the problem of splitting edges by tracking edges from a fine-to-coarse resolution. His algorithm detects edges using the Canny method of NMS of the magnitude of CES-506, University of Essex, U... |

9 |
The fast multilevel fuzzy edge detection of blurry
- Wu, Yin, et al.
- 2007
(Show Context)
Citation Context ...t. The advantage of this technique is that it exploits the estimated local signal characteristics and does not require any overall thresholding procedure. 4.7 Machine Learning Based Methods Wu et al. =-=[20]-=- introduced a fast multilevel fuzzy edge detection algorithm that realizes the fast and accurate detection of the edges from the blurry images. The algorithm first enhances the image contrast by means... |

9 |
Wavelet-based solution to anisotropic diffusion equation for edge detection
- Fontaine, Basu
- 1998
(Show Context)
Citation Context ...unately, under practical situations, this phenomenon is hardly observed. It is an experimental fact that reasonable discretizations of the Perona-Malik equation are rarely unstable. Fontaine and Basu =-=[38]-=- suggest the use of wavelets to solve the anistropic diffusion equation, which will be discussed in the next sub-Section. 4.5 Wavelet Based Methods As it was mentioned, analysing an image at different... |

8 |
Neuro-fuzzy synergism to the intelligent system for edge detection and enhancement
- Lu, Wang, et al.
(Show Context)
Citation Context ...m the image, which leads to its better performance than the Sobel operator, Canny operator, traditional fuzzy edge detection algorithm, and other multilevel fuzzy edge detection algorithms. Lu et al. =-=[9]-=- proposed a fuzzy neural network system for edge detection and enhancement by recovering missing edges and eliminating false edges caused by noise. The algorithm was comprised of three stages, namely,... |

8 |
A.P.Dhawan,"edge detection using a Hopfield neural network", optical engineering 33
- Chao
- 1994
(Show Context)
Citation Context ...onverge. Our experimental results show that the CHNN can obtain more appropriate, more continued edge points than Laplacian-based, Marr-Hildreth 's, Canny's, and waveletbased methods. Chao and Dhawan =-=[23]-=- presented an edge detection algorithm using Hopfield neural network. This algorithm brings up a concept which is different from those conventional differentiation operators, such as Sobel and Laplaci... |

7 |
A new edge detection approach based on image context analysis
- Yu, Chang
(Show Context)
Citation Context ...is usually employed to find the edges. In the last stage, an edge localization process is performed to identify the genuine edges, which are distinguished from those similar responses caused by noise =-=[15]-=-. This paper reviews some dominant literature published in recent two decades on edge detection, including background, significant works, categories and evaluation. Section I is the introduction. Sect... |

6 |
Combined edge detection using wavelet transform and signal registration
- Heric, Zazula
- 2007
(Show Context)
Citation Context ..., e.g. edges, with a zooming procedure that progressively reduces the scale parameter. In this way, coarse and fine signal structures are simultaneously analysed at different scales. Heric and Zazula =-=[13]-=- presented an edge detection algorithm using Haar wavelet transform. They chose Haar wavelet as the mother wavelet function, because it was orthogonal, compact and without spatial shifting in the tran... |

6 |
Contextual-based Hopfield neural network for medical image edge detection,‖
- Chang
- 2006
(Show Context)
Citation Context ...a fuzzy classifier that differentiates image pixels into six classes consisting of background (no edge), speckle (noisy) edge, and four types of edges (in four directions) as shown in Figure 9. Chang =-=[22]-=- employed a special design of neural networks for edge detection. He introduced a method called Contextual Hopfield Neural Network (CHNN) for finding the edges of medical CT and MRl images. Different ... |

6 |
Edge Detection from Local Negative Maximum of Second Derivative
- Giraudon
- 1985
(Show Context)
Citation Context ...of order less than or equal to three. Lines occur at pixels having zero-crossings of the first directional derivative taken in the direction that maximizes the second directional derivative. Giraudon =-=[46]-=- proposed an algorithm for detecting a line at a negative local maximum of the second derivative of the image, rather than a zero-crossing of the first derivative. He estimated the second derivative b... |

5 |
A new efficient SVM based edge detection method
- Zheng, Liu, et al.
- 2004
(Show Context)
Citation Context ...) first order derivative of the image is presented as following. Right from the early stages of edge detection, it was recognized that we can use operators of several dimensions. Rosenfeld et al. [8]-=-=[10]-=- proposed an algorithm to detect edges, commonly known as ―difference of boxes‖, that relies on the use of pairs of neighbourhoods (one neighbourhood on each side of the point under analysis) of sever... |

4 |
On edge focusing‖, Image Vis
- Goshtasby
- 1994
(Show Context)
Citation Context ...illiams and Shah specify the number of scales to be used and the relationship between these scales, they did not suggest the best way to choose the value of σ and under what conditions [5]. Goshtasby =-=[33]-=- proposes an algorithm that works on a modified scale-space representation of an image. The author creates a representation of an image by recording the signs of pixels (instead of the zero-crossings)... |

4 |
et.al, Adaptive determination of filter scales for edge-detection
- Jeong
- 1992
(Show Context)
Citation Context ...he need for a considerable amount memory to store the three-dimensional (3D) edge images [5]. To avoid the common problems associated with integrating edges detected at multiple scales, Jeong and Kim =-=[34]-=- proposed a scheme which automatically determines the optimal scales for each pixel before detecting the final edge map. To find the optimal scales for a Gaussian filter, they define an energy functio... |

4 |
Optimal Parameters for Edge Detection. In
- Bennamoun, Boashash, et al.
- 1995
(Show Context)
Citation Context ...hm is that it assumes the noise is Gaussian with known variance. In practical situations, however, the noise variance has to be estimated. The algorithm is also very computationally intensive [5]. In =-=[36]-=-, Bennamoun et al. present a hybrid detector that divides the tasks of edge localization and noise suppression between two sub-detectors. This detector is the combination of the outputs from the Gradi... |

4 |
Line Detection Using an Optimal IIR
- Ziou
- 1991
(Show Context)
Citation Context ...in problem with Giraudon's detector [46] comes from the use of the gradient since at the peak point, the gradient value is too small to be used. Using a 1D ideal roof model and Canny's criteria, Ziou =-=[47]-=- derived an optimal line detector. Koundinya and Chanda [48] proposed an algorithm based on combinatorial search. The basic idea behind this algorithm is to locate lines that maximize an ad-hoc confid... |

3 |
Non-linear fourth-order image interpolation for subpixel edge detection and localization
- Hermosilla, Bermejo, et al.
(Show Context)
Citation Context ...tion) first order derivative of the image is presented as following. Right from the early stages of edge detection, it was recognized that we can use operators of several dimensions. Rosenfeld et al. =-=[8]-=--[10] proposed an algorithm to detect edges, commonly known as ―difference of boxes‖, that relies on the use of pairs of neighbourhoods (one neighbourhood on each side of the point under analysis) of ... |

3 |
A wavelet based multi resolution edge detection and tracking
- Shih, Tseng
- 2005
(Show Context)
Citation Context ...t different characteristics, for example, it is not likely that they contain linked pieces of the same edge. They used the sum of squared differences (SSD) as a measure for uniformity. Shih and Tseng =-=[14]-=- combined a gradient-based edge detection and a wavelet based multiscale edge tracking to extract edges. The proposed contextual filter detects edges from the finest scale gradient images and then, th... |

3 |
A Bayesian Approach to Edge Detection in Noisy Images
- Santis, Sinisgalli
- 1999
(Show Context)
Citation Context ...ity of Essex, U.K. Page 25blending functions is the key to using models based on computational learning algorithms such as neural networks and fuzzy systems for edge detection. Santis and Sinisgalli =-=[19]-=- proposed a statistical edge detection algorithm using a linear stochastic signal model derived from a physical image description. The presence of an edge was modelled as a sharp local variation of th... |

2 |
A review on edge detection based on filtering and differentiation
- Pinho, Almeida
- 1997
(Show Context)
Citation Context ...ed on the Gaussian function. This is not surprising since the Gaussian has been emerging as a very important function in several areas of image processing and, specially, in multi-resolution analysis =-=[3]-=-. The core of image differentiation is mainly based on discrete convolution that estimates image derivatives either by the gradient or Laplacian. The edges are localized either by local maxima on the ... |

2 |
Detection andClassification of Edgesin Color
- Koschan, Abidi
- 2005
(Show Context)
Citation Context ...ation operator. Consequently, it is not easy to find a single threshold value for a given image. An automatic rule to compute the threshold for the Laplacian of Gaussian detector has been proposed in =-=[18]-=-. This rule is empirical, no justification has been given and it has been tested only on synthetic data. It is also proposed a cleaning rule for multi-scale edge detection based on the behaviour of th... |

2 | Use Of Predictive Coding Distribution For Edge Detection
- Cramariuc, Tabug, et al.
(Show Context)
Citation Context ... on the output level of the converged network can be easily set up at 0.5 level to extract edges. The experimental results are presented to show the effectiveness and capability of this algorithm. In =-=[39]-=-, the authors analyzed how the predictive distribution estimated using a set of context dependent nonlinear adaptive predictors can be used to localize edges in an images. Since the adaptive predictor... |

2 |
Edge Detection Using Dynamic Optimal Partitioning
- Scargle, Quweider
(Show Context)
Citation Context ...s those characteristic to certain textures, the proposed predictive edge detection scheme can be a practical way to conceal the relative high contrast of certain texture regions. Scargle and Quweider =-=[40]-=- proposed an edge detector based on finding major change points in a local 1D window of the image intensity values of the rows or columns. The approach amounts to separating the pixels in the window i... |

2 |
Edge Evaluation Using Necessary Comp onents
- Venkatesh, Kitchen
- 1992
(Show Context)
Citation Context ...cult to determine the type of error committed by the detector. Kitchen and Rosenfeld's measure [42] combines errors that arise due to an edge's thickness and lack of continuity. Venkatesh and Kitchen =-=[43]-=- used four error types which reflect the major difficulties encountered in edge detection: non-detection of true edges, detection of false edges, detection of several edges instead of an edge one pixe... |

2 |
and B.Chanda.Detecting Lines in Gray Level Images Using Search Techniques
- Koundinya
- 1994
(Show Context)
Citation Context ...of the gradient since at the peak point, the gradient value is too small to be used. Using a 1D ideal roof model and Canny's criteria, Ziou [47] derived an optimal line detector. Koundinya and Chanda =-=[48]-=- proposed an algorithm based on combinatorial search. The basic idea behind this algorithm is to locate lines that maximize an ad-hoc confidence measure. The confidence measure of a candidate pixel is... |