(IJCSIS) International Journal of Computer Science and Information Security,Vol. 9, No. 2, February 2011
Image Processing: The Comparison of the EdgeDetection Algorithms for Images in Matlab
Ehsan Azimirad
Department of electrical and computer engineering,Tarbiat Moallem University of Sabzevar,Sabzevar, Iraneazimi@sttu.ac.ir
Javad Haddadnia
Department of electrical and computer engineering,Faculty of Electrical Collage, Tarbiat Moallem Universityof Sabzevar, Sabzevar, Iranhaddadnia@sttu.ac.ir
Abstract
Edge detection is the first step in image segmentation.Image Segmentation is the process of partitioning a digital imageinto multiple regions or sets of pixels. Edge detection is one of themost frequently used techniques in digital image processing. Thegoal of edge detection is to locate the pixels in the image thatcorrespond to the edges of the objects seen in the image. Filtering,Enhancement and Detection are three steps of Edge detection.Images are often corrupted by random variations in intensityvalues, called noise. Some common types of noise are salt andpepper noise, impulse noise and Gaussian noise. However, thereis a trade-off between edge strength and noise reduction. Morefiltering to reduce noise results in a loss of edge strength. In orderto facilitate the detection of edges, it is essential to determinechanges in intensity in the neighborhood of a point. Enhancementemphasizes pixels where there is a significant change in localintensity values and is usually performed by computing thegradient magnitude. Many points in an image have a nonzerovalue for the gradient, and not all of these points are edges for aparticular application. Therefore, some method should be used todetermine which points are edge points. Four most frequentlyused edge detection methods are used for comparison. These are:Roberts Edge Detection, Sobel Edge Detection, Prewitt EdgeDetection and Canny Edge Detection. One the other method inedge detection is spatial filtering. This Paper represent a specialmask for spatial filtering and compare throughput the standardedge detection algorithms (Sobel, Canny, Prewit & Roberts) withthe spatial filtering.
Keywords
-Spatial Filtering, Median Filter, Edge Detection, ImageSegmentation.
I.
I
NTRODUCTION
Over the years, several methods have been proposed for theimage edge detection which is the method of marking points ina digital image where luminous intensity changes sharply forwhich different type of methodology have been implementedin various applications like traffic speed estimation [5], Imagecompression [6], and classification of images [7]. Most of thetraditional edge-detection algorithms in image processingtypically convolute a filter operator and the input image, andthen map overlapping input image regions to output signalswhich lead to considerable loss in edge detection [8,9].Edge and feature points are basic low level primitives forimage processing. Edge and feature detection are two of themost common operations in image analysis. An edge in animage is a contour across which the brightness of the imagechanges abruptly. In image processing, an edge is ofteninterpreted as one class of singularities. In a function,Singularities can be characterized easily as discontinuitieswhere the gradient approaches Infinity. However, image datais discrete, so edges in an image often are defined as the Localmaxima of the gradient. This is the definition we will use here.Operations in image processing, This topic has attracted manyresearchers and many achievements have been made [11-18].For Such as: Rooms et al proposed to estimate the out-of focus blur in wavelet domain by examining the sharpness of the sharpest edges [11]. Hanghang Tong et al proposed newblur detection schemes which can determine whether an imageis blurred or not and to what extent an image is blurred. Whichraises the demand for image quality assessment in terms of blur Based on the edge type and sharpness analysis using Harrwavelet transforms [12]. X. Marichal, proposed using DCTinformation to qualitatively characterize blur extent [13]Berthold K., ET AL describes the processing performed in thecourse of producing a line drawing from an image obtainedthrough an image dissector camera. The edgemarking phaseuses a non-linear parallel line-follower [14]. Lixia Xue et alproposed An edge detection algorithm for multispectralremote sensing image, they extended the onedimensionalcloud-space mapping model to the multidimensional model[15].Mike Heath etal, presented a paradigm based onxperimental psychology and statistics, in which humans ratethe output of low level vision algorithms. They demonstratethe proposed experimental strategy by comparing four well-known edge detectors: Canny, Nalwa–Binford, Sarkar–Boyer,and Sobel [16], Hoover
etal
at USF have recently conductedsuch a comparison study based on manually constructedground truth for range segmentation tasks [17]. Krishna KantChintalapudi et alshowed that such localized edge detectiontechniques are non-trivial to design in an arbitrarily deployedsensor network. They defined the notion of an edge anddevelop performance metrics for evaluating localized edgedetection algorithms[10,18].Usage of specific linear time-invariant (LTI) filters is themost common procedure applied to the edge detectionproblem, and the one which results in the least computational
108 http://sites.google.com/site/ijcsis/ISSN 1947-5500
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 9, No. 2, February 2011
effort. In the case of first-order filters, an edge is interpreted asan abrupt variation in gray level between two neighbor pixels.The goal in this case is to determine in which points in theimage the first derivative of the gray level as a function of position is of high magnitude. By applying the threshold to thenew output image, edges in arbitrary directions are detected.In other ways the output of the edge detection filter is theinput of the polygonal approximation technique to extractfeatures which to be measured. A very important role is playedin image analysis by what are termed feature points, pixelsthat are identified as having a special property. Feature pointsinclude edge pixels as determined by the well-known classicedge detectors of PreWitt, Sobel, Roberts, Canny and SpatialFiltering. Classical operators identify a pixel as a particularclass of feature point by carrying out some series of operationswithin a window centered on the pixel under scrutiny. Theclassic operators work well in circumstances where the area of the image under study is of high contrast. In fact, classicoperators work very well within regions of an image that canbe simply converted into a binary image by simplethresholding[1].This paper is organized as follows. Section II is for thepurpose of providing some information about edge detection.Section III is focused on simulation results and also focusedon comparison of various Edge Detection Methods. Section IVpresents the conclusion.II.
E
DGE
D
ETECTION
Edge detection techniques transform images to edge imagesbenefiting from the changes of grey tones in the images. Edgesare the sign of lack of continuity, and ending. As a result of this transformation, edge image is obtained withoutencountering any changes in physical qualities of the mainimage. Objects consist of numerous parts of different colorlevels. In an image with different grey levels, despite anobvious change in the grey levels of the object, the shape of the image can be distinguished in Fig.1.
Figure 1. Type of Edges (a) Step Edge (b) Ramp Edge (c) Line Edge (d)Roof Edge
An Edge in an image is a significant local change in theimage intensity, usually associated with a discontinuity ineither the image intensity or the first derivative of the imageintensity. Discontinuities in the image intensity can be eitherStep edge, where the image intensity abruptly changes fromone value on one side of the discontinuity to a different valueon the opposite side, or Line Edges, where the image intensityabruptly changes value but then returns to the starting valuewithin some short distance. However, Step and Line edges arerare in real images. Because of low frequency components orthe smoothing introduced by most sensing devices, sharpdiscontinuities rarely exist in real signals. Step edges becomeRamp Edges and Line Edges become Roof edges, whereintensity changes are not instantaneous but occur over a finitedistance. Illustrations of these edge shapes are shown in Fig.1.
A.
Steps in Edge Detection
Edge detection contain three steps namely Filtering,Enhancement and Detection. The overview of the steps inedge detection are as follows.
1) Filtering:
Images are often corrupted by randomvariations in intensity values, called noise. Some commontypes of noise are salt and pepper noise, impulse noise andGaussian noise. Salt and pepper noise contains randomoccurrences of both black and white intensity values.However, there is a trade-off between edge strength and noisereduction. More filtering to reduce noise results in a loss of edge strength.
2) Enhancement:
In order to facilitate the detection of edges,it is essential to determine changes in intensity in theneighborhood of a point. Enhancement emphasizes pixelswhere there is a significant change in local intensity valuesand is usually performed by computing the gradientmagnitude.
3) Detection:
Many points in an image have a nonzero valuefor the gradient, and not all of these points are edges for aparticular application. Therefore, some method should be usedto determine which points are edge points. Frequently,thresholding provides the criterion used for detection.
B.
Edge Detection Methods
Three most frequently used edge detection methods are usedfor comparison. These are (1) Roberts Edge Detection, (2)Sobel Edge Detection, (3) Prewitt edge detection and (4)Canny edge detection. One the other method in edge detectionis spatial filtering. The details of methods as follows:
1) The Roberts Detection:
The Roberts Cross operatorperforms a simple, quick to compute, 2-D spatial gradientmeasurement on an image. It thus highlights regions of highspatial frequency which often correspond to edges. In its mostcommon usage, the input to the operator is a grayscale image,as is the output. Pixel values at each point in the outputrepresent the estimated absolute magnitude of the spatialgradient of the input image at that point. Fig.2. shows RobertsMask.
109 http://sites.google.com/site/ijcsis/ISSN 1947-5500
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 9, No. 2, February 2011
Figure 2. Roberts Mask
2) The Prewitt Detection:
The prewitt edge detector is anappropriate way to estimate the magnitude and orientation of an edge. Although differential gradient edge detection needs arather time consuming calculation to estimate the orientationfrom the magnitudes in the x and y-directions, the compassedge detection obtains the orientation directly from the kernelwith the maximum response. The prewitt operator is limited to8 possible orientations, however experience shows that mostdirect orientation estimates are not much more accurate. Thisgradient based edge detector is estimated in the 3x3neighbourhood for eight directions. All the eight convolutionmasks are calculated. One convolution mask is then selected,namely that with the largest module. Fig.3. shows PrewittMask.
Figure 3. Prewitt Mask
3) The Sobel Detection:
The Sobel operator performs a 2-Dspatial gradient measurement on an image and so emphasizesregions of high spatial frequency that correspond to edges.Typically it is used to find the approximate absolute gradientmagnitude at each point in an input grayscale image. In theoryat least, the operator consists of a pair of 3x3 convolutionkernels as shown in Figure 4. One kernel is simply the otherrotated by 90o.This is very similar to the Roberts Crossoperator. The convolution masks of the Sobel detector aregiven in Fig.4. Fig.5. shows Edge patterns for Sobel edgedetector.
Figure 4. Sobel MaskFigure 5. Edge patterns for Sobel edge detector
4) The Canny Detection:
Canny edge detection is animportant step towards mathematically solving edge detectionproblems. This edge detection method is optimal for stepedges corrupted by white noise. Edge detection with lowprobability of missing true edges, and a low probability of detecting false edges. [2] The Canny algorithm uses anoptimal edge detector based on a set of criteria which includefinding the most edges by minimizing the error rate, markingedges as closely as possible to the actual edges to maximizelocalization, and marking edges only once when a single edgeexists for minimal response.[3]Canny used three criteria to design his edge detector. Thefirst requirement is reliable detection of edges with lowprobability of missing true edges, and a low probability of detecting false edges. Second, the detected edges should beclose to the true location of the edge. Lastly, there should beonly one response to a single edge. To quantify these criteria,the following functions are defined:
01022
( )( ) .( )
f x dxASNR f nf x dx
−∞∞−∞
=
∫ ∫
(1)
1022
(0)( ) .( )
f ASNR f nf x dx
∞−∞
′=
′
∫
(2)where A is the amplitude of the signal and n20 is thevariance of noise. SNR(f) defines the signal-to-noise ratio andLoc(f) defines the localization of the filter f(x).The Canny edge detection algorithm runs in 5 separate steps:1. Smoothing: Blurring of the image to remove noise.2. Finding gradients: The edges should be marked where thegradients of the image has large magnitudes.3. Non-maximum suppression: Only local maxima shouldbe marked as edges.4. Double thresholding: Potential edges are determined bythresholding.5. Edge tracking by hysteresis: Final edges are determinedby suppressing all edges that are not connected to a verycertain (strong) edge.[19]
5) The Spatial Filtering Detection:
we implement imageedge detection so that we can identify the boundary of object
110 http://sites.google.com/site/ijcsis/ISSN 1947-5500
Search History:
Result 00 of 00
00 results for result for
p.
Image Processing: The Comparison of the Edge Detection Algorithms for Images in Matlab
Edge detection is the first step in image segmentation. Image Segmentation is the process of partitioning a digital image into multiple regions or sets of pixels. Edge detection is one of the most frequently used techniques…
(More)
2.1K
Reads
2
Readcasts
14
Embed Views
Recommended
More From This User
Notes
Load more
