Docstoc

Skin Lesion Segmentation Algorithms using Edge Detectors

Document Sample
Skin Lesion Segmentation Algorithms using Edge Detectors Powered By Docstoc
					                                                           (IJCSIS) International Journal of Computer Science and Information Security,
                                                           Vol. 9, No. 5, May 2011




     Skin Lesion Segmentation Algorithms using Edge
                       Detectors

                  J.H.Jaseema Yasmin1                                                            M.Mohamed Sathik2
  Associate Professor, Department of Computer Science                              Associate Professor in Computer Science
                     and Engineering                                            Sadakathullah Appa College, Tirunelveli – India
   National College of Engineering,Tirunelveli, India.                                      mmdsadiq@gmail.com
                 jaseemay@yahoo.co.in


Abstract— An effective segmentation algorithm using log edge               clinician. ABCD rule is one of the most widely used methods
detector, for border detection of real skin lesions is presented           for evaluating pigmented skin lesions with the naked-eye [7].
which insinuate the excessive growth or regression of a                    When the pigmented skin lesions are small or/and regular in
melanoma, that helps in early detection of malignant melanoma              shape or color, however, this system may fail sometimes[4].
and its performance is compared with the segmentation                      The most hastily increasing cancer in the world is malignant
algorithm using canny detector, developed by us previously for
                                                                           melanoma. Since melanoma can be cured with a simple
border detection of real skin lesions. The experimental results
demonstrate the successful border detection of noisy real skin             expurgation if detected early, early diagnosis is particularly
lesions by the effective segmentation algorithm using log detector.        important [5].
We conclude that the segmentation algorithm using log detector,                      Automated border detection is vital for the image
segments the lesion from the image even in the presence of noise           analysis because the border structure provides important
for a variety of lesions, and skin types and its performance is            information for precise diagnosis, as many clinical features
better than the segmentation algorithm that we have developed              such as asymmetry, border irregularity, and abrupt border
previously that uses canny detector, for border detection of real          cutoff are calculated directly from the border.
skin lesions for noisy skin lesion diagnosis.                                        Automated border detection is a exigent task due to
    Keywords- Segmentation; Skin Lesion; log edge detector; canny
                                                                           the following reasons: low contrast between the lesion and the
edge detector; Border detection; Melanoma.
                                                                           surrounding skin, irregular and fuzzy lesion borders, features
                                                                           such as skin lines ,blood vessels , hairs , and air bubbles,
                       I.    INTRODUCTION                                  variegated coloring inside the lesion , and fragmentation due to
         Image segmentation is used to locate objects and                  various reasons such as scar-like depigmentation [5].
boundaries in images, is not a simple task due to the great                          To considerably reduces morbidity and mortality,
variety of lesions, skin types, presence of hair etc [14].                 detection of malignant melanoma should be done in its early
          Once a image is selected , the system should provide             stages. We can also hoard hundreds of millions of dollars by
an automatic identification (or segmentation) of the lesion,               early detection that otherwise would be spent on the treatment
which aims at identifying the lesion and separate it from the              of advanced diseases. There is a very high likelihood that the
background. The algorithm will have to be able to eradicate                patient will survive, if cutaneous melanoma is detected in its
noise and other undesired features in the image, and to                    early stages and removed. The ABCDs of melanoma are [3]:
correctly segment the lesion [1]. Visual segmentation of tumor             asymmetry, border irregularity, color variegation, and diameter
by dermatologist is simple in most of the cases. When                      greater than 6 mm. Image analysis techniques for measuring
transition between lesion and surrounding skin is too smooth,              these features have been developed. Measurement of image
sporadically some irreducible fuzziness remains. The copious               features for diagnosis of melanoma requires that first the
papers on boundary detection of skin tumors expound that it is             lesions be detected and localized in an image. It is essential that
still an open dilemma for computers. As a matter of fact,                  lesion boundaries are determined accurately so that
lesions show a discrepancy in size, color, texture [2].                    measurements, e.g. maximum diameter, asymmetry,
         The process of contour extraction of different objects            irregularity of the boundary, and color characteristics can be
from background is edge detection and it is very imperative                precisely computed. Various image segmentation methods have
to image understanding and computer vision. Problems with                  been developed for delineating lesion boundaries [6].
edge detection are edge location errors, false edges, and broken
or missing edge fragments [3].
          To reduce mortality early detection and surgical                                 II.     REVIEW OF RELATED WORK
excision is currently the only approach, because advanced skin                       Due to the great variety of lesions, skin types,
cancers remain incurable. The conventional screening tests                 presence of hair and so forth, the segmentation stage is not a
require a skin naked-eye examination by an experienced                     straightforward task. A variety of image segmentation methods



                                                                      30                               http://sites.google.com/site/ijcsis/
                                                                                                       ISSN 1947-5500
                                                          (IJCSIS) International Journal of Computer Science and Information Security,
                                                          Vol. 9, No. 5, May 2011


have been proposed for this purpose. L. Xu et al. developed a                             III.   PROPOSED METHODOLOGY
three-step segmentation method using the properties of skin                  The image segmentation algorithm using Log Edge
cancer images. The steps of their method are as follows: 1.               Detector, for border detection of skin lesions, developed by us
Preprocessing: a color image is first transformed into an                 [18], that reveals the global structure irregularity, which may
intensity image in such a way that the intensity at a pixel               evoke excessive cell growth or regression of a melanoma is
shows the color distance of that pixel with the color of the              discussed in this paper. This algorithm is applied to the image
background. The color of the background is taken to be the                containing the lesion
median color of pixels in small windows in the four corners of
the image. 2. Initial segmentation: a threshold value is                  A. Image Segmentation Algorithm using log edge detector
determined from the average intensity of high gradient pixels in            Step 1 : The RGB image is converted to grayscale image
the obtained intensity image. This threshold value is used to               Step 2 : Salt and pepper noise is added to the grayscale
find approximate lesion boundaries. 3. Region refinement: a                           image .The noisy image is the input image.
region boundary is refined using edge information in the image.             Step 3:Median filter used as the background noise reduction
This involves initializing a closed elastic curve at the                             technique to filter noise.
approximate boundary, and shrinking and expanding it to fit to              Step 4 : After noise reduction, the image is converted to a
the edges in its neighbourhood [6].                                                  black and white image, based on threshold,
        We have previously developed a segmentation                         Step 5: The black and white image got is converted into xor
algorithm[17], to extract the true border that reveals the global                    image
structure irregularity, which may evoke excessive cell growth               Step 6:The Log Edge detector is used to find the edges in the
or regression of a melanoma. The steps of this algorithm[17]                         xor image .We get the edge detected image.
are as follows: 1.This algorithm is applied to the input image              Step 7 : The pixel on the border of the object is found. To
containing the lesion, where the input RGB image is converted                        find the pixel on the border of the object (Lesion)
to grayscale image. 2. Salt and pepper noise is added to the                         the binary image is used to find the row co-ordinate
grayscale image and background noise reduction techniques are                        of the pixel on the border of the object and the edge
used to filter noise. 3.The noise filtered image is converted to a                   detected image is used to find the column co-
binary image, based on threshold. 4. Then the binary image is                        ordinate of the pixel on the border of the object to
converted to xor image 5. The Canny Edge detector is used to                         be traced
find the edges in the xor image .We get the edge detected                   Step 8 : Using this pixel found on the border of the object
image.6. The pixel on the border of the object is found.7. Using                     (Lesion) as the starting pixel , the border of the
this pixel found on the border of the object (Lesion) as the                         lesion is traced using the robust segmentation
starting pixel, the border of the lesion is traced, using the                        algorithm[18] using log detector, successfully .
segmentation algorithm[17] using canny detector .
       Image segmentation is conceivably, the most                        B.    Median filtering
premeditated area in computer vision, with copious methods                     To reduce "salt and pepper" noise, median filtering is a
reported. A segmentation method is usually designed taking                nonlinear operation often used in image processing. Median
into consideration the properties of a particular class of images.        filtering is more effective than convolution when the goal is to
The algorithm will have to be able to confiscate noise and other          simultaneously reduce noise and preserve edges.
undesired features in the image, and to correctly segment the             C. Edge Detection
lesion. Developing robust and proficient algorithm for medical
image segmentation has been a exigent area of interesting                     An edge is a set of connected pixels that lie on the
research interest, over the last decade [15].                             boundary between two regions[10]. An image can be
       The medical images generally are bound to restrain                 segmented by detecting those discontinuities.
noise while acquisition. An efficient and robust segmentation                 The key to a satisfactory segmentation result lies in keeping
algorithm against noise is needed for medical image                       a balance between detecting accuracy and noise immunity. If
segmentation. Accurate segmentation of medical images is                  the level of detecting accuracy is too high, noise may bring in
therefore highly challenging, however, accurate segmentation              fake edges making the outline of images unreasonable
of these images is imperative in correct diagnosis by clinical            .Otherwise,         some parts of the image outline may get
tools [16].                                                               undetected and the position of objects may be mistaken if the
        In this paper, we have compared the performance of                degree of noise immunity is excessive [12].
robust segmentation algorithm using log detector for border                     Edge detection is a most common approach for detecting
detection of real skin lesions for noisy skin lesion images               meaningful discontinuities in grey level .Such discontinuities
developed by us[18], with the segmentation algorithm using                are detected using first order and second order derivatives [5].
canny detector for border detection of real skin lesions for              The first order derivative of choice is the gradient. The gradient
noisy skin lesion images developed by us[17] .                            of the 2D function f(x, y), is defined as a vector. The
                                                                          magnitude of this vector is given by

                                                                                         g = [Gx2+Gy2]1/2                              (1)




                                                                     31                               http://sites.google.com/site/ijcsis/
                                                                                                      ISSN 1947-5500
                                                           (IJCSIS) International Journal of Computer Science and Information Security,
                                                           Vol. 9, No. 5, May 2011


Where Gx = ∂ƒ/∂x and             Gy= ∂ƒ/∂y                                  J.H.Jaseema Yasmin et al. [17], when they are applied to the
      The second derivative in image processing is computed                 different types of skin lesions, with noise.
using the laplacian. The laplacian is soldem used by itself for             For the different types of Skin Lesions taken , J.H.Jaseema
edge detection because as a second order derivative it is                   Yasmin et al. [17] method poorly delineates the boundary for
unacceptably sensitive to noise, its magnitude produces double              some of the skin lesions. The Figure 1(c), 2(c), 3(c)
edges and it is unable to detect edge direction. However                    demonstrates the failure of this method[17] to delineate the
Laplacian can be a powerful complement when used in                         boundary of the lesion of various types.
combination of other edge detection techniques. The basic idea
behind edge detection is to find places in an image where the               The robust segmentation algorithm using log detector[18],
intensity changes very rapidly using one of the two general                 converts the original skin lesion image (skin lesion 1-9) in
                                                                            Figure 1(a), 2(a), 3(a), 4(a), 5(a), 6(a), 7(a), 8(a) and 9(a) into a
criteria:
                                                                            gray scale image. 20% salt and pepper noise was added to the
   1. Find places where the first derivative of the intensity is
                                                                            original image and that is illustrated in Figure 1(b), 2(b), 3(b),
greater in magnitude than a specified threshold.                            4(b), 5(b), 6(b), 7(b), 8(b) and 9(b).The noisy image is the
   2. Find places where the second derivative of the intensity              input image to the proposed algorithm. The median filter is
has zero crossing.                                                          applied and the noise is removed. After noise removal the
    1)Laplacian of Gaussian Detector : Consider the Gaussian                image is enhanced. Based on a threshold value the enhanced
function                                                                    image is converted to black and white image. This
          h(r) = -℮ - r2/2σ2                              (2)               algorithm[18] converts the black and white image into xor
                                                                            image and the edges are detected using log edge detector. The
     Where r2=x2+y2 and σ is the standard deviation. This is a              black and white image is used to find the row co-ordinate of the
smoothing function, which if convolved with an image, will                  pixel on the border of the object and the edge detected image
                                                                            is used to find the column co-ordinate of the pixel on the border
blur it. The degree of blurring is determined with the value of
                                                                            of the object to be traced and using this pixel found on the
σ. The Laplacian of this function (the second derivative with
                                                                            border of the object as the starting pixel , the border of the
respect to r) is ( - [(r2-σ2) /σ4] ℮ -r2 / 2σ2 )                            lesion is traced using the robust segmentation algorithm[18]
     This function is called Laplacian of Gaussian. Because the             successfully is shown in Figure 1(d), 2(d). 3(d), 4(d), 5(d), 6(d),
second derivative is a linear operation, convolving the image               7(d), 8(d) and 9(d). The robust segmentation algorithm using
with the above said function, is the same as convolving the                 log detector[18], segments the lesion from the image even in
image with the smoothing function first and then computing the              the presence of noise and presence of hair for a variety of
Laplacian of the result. This is the key concept underlying the             lesions, and skin types.
LOG detector. The LOG detector finds the edges by looking
for zero crossing after filtering f(x, y) with a Gaussian filter
[11]
                 IV.    RESULTS AND DISCUSSION
   An image segmentation algorithm to extract the true border
of the skin lesions, that is helpful in the diagnosis of
melanoma, has been implemented using Matlab. Our aim is to
select an image and the system should impart an automatic
identification (or segmentation) of the lesion, which aims at
identifying the lesion and separate it from the background. The
algorithm will have to be able to remove noise and other
undesired features in the image, and to correctly segment the
lesion. The algorithm should work well even when the
transition between lesion and surrounding skin is too smooth.
The segmentation stage is not a candid task due to the great
variety of lesions, skin types, presence of hair etc. The
segmentation algorithm using log detector[18], works well
even in the presence of noise and hair, to detect the border of
the lesion, which helps the medical practitioners in diagnosis.

   The robust segmentation algorithm using log detector for
border detection of real skin lesions[18] was applied to a                   Figure 1.Demonstration of border detection for Skin lesion 1
variety of skin lesions, and skin types. Figure 1(a), 2(a), 3(a),           (a) Skin lesion (b) Noisy image (c) Border traced image by robust
4(a), 5(a), 6(a), 7(a) ,8(a) and 9(a) illustrates different types of        segmentation algorithm using canny detector[17] for a noisy image. (d)
                                                                            Border traced image by robust segmentation algorithm using LOG detector for
original skin lesions. Figure 1(c), 2(c), 3(c), 4(c), 5(c), 6(c),           a noisy image.
7(c), 8(c) and 9(c) shows the final output results of the                     Figure 1(a) referred from L.Xua et.al [6]
segmentation algorithm using canny edge detector[17], by



                                                                       32                                   http://sites.google.com/site/ijcsis/
                                                                                                            ISSN 1947-5500
                                                                  (IJCSIS) International Journal of Computer Science and Information Security,
                                                                  Vol. 9, No. 5, May 2011




Figure 2.Demonstration of border detection for Skin lesion 2
(a) Skin lesion (b) Noisy image (c) Border traced image by robust                  Figure 4.Demonstration of border detection for Skin lesion 4
segmentation algorithm using canny detector[17] for a noisy image. (d)             (a) Skin lesion (b) Noisy image (c) Border traced image by robust
Border traced image by robust segmentation algorithm using LOG detector for        segmentation algorithm using canny detector[17] for a noisy image. (d)
a noisy image.                                                                     Border traced image by robust segmentation algorithm using LOG detector for
Figure 2(a) referred from L.Xua et.al [6]                                          a noisy image.
                                                                                   Figure 4(a) referred from L.Xua et.al [6]




Figure 3.Demonstration of border detection for Skin lesion 2                       Figure 5.Demonstration of border detection for Skin lesion 5
(a) Skin lesion (b) Noisy image (c) Border traced image by robust                  (a) Skin lesion (b) Noisy image (c) Border traced image by robust
segmentation algorithm using canny detector[17] for a noisy image. (d)             segmentation algorithm using canny detector[17] for a noisy image. (d)
Border traced image by robust segmentation algorithm using LOG detector for        Border traced image by robust segmentation algorithm using LOG detector for
a noisy image.                                                                     a noisy image.
Figure 3(a) referred from M.Emre Celebia et.al [5]                                 Figure 5(a) referred from L.Xua et.al [6]




                                                                              33                                   http://sites.google.com/site/ijcsis/
                                                                                                                   ISSN 1947-5500
                                                                  (IJCSIS) International Journal of Computer Science and Information Security,
                                                                  Vol. 9, No. 5, May 2011




 Figure 6.Demonstration of border detection for Skin lesion 6
                                                                                   Figure 8.Demonstration of border detection for Skin lesion 8
 (a) Skin lesion (b) Noisy image (c) Border traced image by robust
segmentation algorithm using canny detector[17] for a noisy image. (d)             (a) Skin lesion (b) Noisy image (c) Border traced image by robust
Border traced image by robust segmentation algorithm using LOG detector for        segmentation algorithm using canny detector[17] for a noisy image. (d)
a noisy image.                                                                     Border traced image by robust segmentation algorithm using LOG detector for
                                                                                   a noisy image.
Figure 6(a) referred from L.Xua et.al [6]
                                                                                   Figure 8(a) referred from L.Xua et.al [6]




Figure 7.Demonstration of border detection for Skin lesion 7                       Figure 9.Demonstration of border detection for Skin lesion 9
 (a) Skin lesion (b) Noisy image (c) Border traced image by robust                 (a) Skin lesion (b) Noisy image (c) Border traced image by robust
segmentation algorithm using canny detector[17] for a noisy image. (d)             segmentation algorithm using canny detector[17] for a noisy image. (d)
Border traced image by robust segmentation algorithm using LOG detector for        Border traced image by robust segmentation algorithm using LOG detector for
a noisy image.                                                                     a noisy image.
Figure 7(a) referred from L.Xua et.al [6]                                          Figure 9(a) referred from L.Xua et.al [6]




                                                                              34                                     http://sites.google.com/site/ijcsis/
                                                                                                                     ISSN 1947-5500
                                                                        (IJCSIS) International Journal of Computer Science and Information Security,
                                                                        Vol. 9, No. 5, May 2011


                                                                                                                    REFERENCES
                                  Performance Chart
                                                                                      [1] Teresa Mendonc¸a, Andr´e R. S. Marc¸al, Angela Vieira Jacinto
                                                                                           C. Nascimento, Margarida Silveira, Jorge S. Marques, Jorge
                                                                                           Rozeira,”Comparison of Segmentation Methods for Automatic
                                                                                           Diagnosis of Dermoscopy Images”,Proc. of the 29th Annual
   Performance




                                                                                           International Conference of the IEEE EMBS Cité Internationale,
                                                                                           Lyon, France,August 23-26, 2007.pp. 6572-6575
                                                             log
                                                                                      [2] Arthur Tenenhaus1, Alex Nkengne2, Jean-François Horn3,4,
                          canny                                                             Camille Serruys3,4, Alain Giron3,4 and Bernard Fertil5
                                                                                            “Detection of melanoma from dermoscopic images of naevi
                                                                                            acquired under uncontrolled conditions”, Author manuscript,
                 Segmentation Algorithms using Log and Canny Edge Detectors
                                                                                            published in Skin Research and Technology Vol. 16, N° 1 (2009)
                                                                                             pp. 85-97
                                                                                      [3] LI Xue-wei, ZHANG Xin-rong, “ A Perceptual Color Edge
                                                                                            Detection Algorithm”, 2008 International Conference on
Figure10. Performance comparision chart of the segmentation algorithm                       Computer Science and Software Engineering,pp.297-300
using LOG edge detector and Canny edge detector for tracing the border of             [4] Alfonso Baldi 1,2,*, Marco Quartulli 3, Raffaele Murace 2,
noisy skin lesion images
                                                                                            Emanuele Dragonetti 2, Mario Manganaro 3, Oscar Guerra 3
                                                                                            and Stefano Bizzi 3,Automated Dermoscopy Image Analysis of
                                                                                             Pigmented Skin Lesions , Cancers 2010, 2, 262-273
     The Performance comparision chart of the segmentation
algorithm using log edge detector[18] and Canny edge                                    [5] M.Emre Celebi a, Hitoshi Iyatomib, Gerald Schaeferc,William V.
detector[17] for tracing the border of noisy skin lesion images                             Stoeckerd, “Lesion border detection in dermoscopy images”,
is shown in Figure 10. The segmentation algorithm which uses                                Computerized Medical Imaging and Graphics 33 (2009) 148–153
canny detector to trace the edges of the skin lesion in a noisy                       [6] L. Xu a, M. Jackowskia, A. Goshtasbya,*, D. Rosemanb, S. Binesb,
skin image fail to detect the edges in some of the cases as                                 C. Yuc, A. Dhawand, A. Huntleye , “Segmentation of skin cancer
shown in Figure 1(c), Figure 2(c), Figure 3(c). The                                          images”, Image and Vision Computing 17 (1999) 65–74
segmentation algorithm which uses log detector to trace the                           [7] Harald Ganster*, Axel Pinz, Reinhard Röhrer, Ernst Wildling,
edges of the skin lesion in a noisy skin image successfully                                 Michael Binder, and Harald Kittler, “Automated Melanoma
detects the edges in all of the cases as shown in Figure 1(d) to                            Recognition ”, IEEE Transactions on Medical Imaging, vol. 20,
Figure 9(d). So the performance of the segmentation algorithm                               no. 3, March 2001
using log edge detector for tracing the border of noisy skin                          [8] Roberto Rodríguez* and Ana G. Suarez, “ A new algorithm for
lesion images is better than the performance of the                                         image segmentation by using iteratively the mean shift
segmentation algorithm using canny edge detector for tracing                                filtering”, Scientific Research and Essay Vol. 1 (2), pp.043-048,
the border of noisy skin lesion images.                                                     November 2006
                                                                                       [9] Wen-Xiong Kang, Qing-Qiang Yang, Run-peng Liang, “The
                                  V.     Conclusion                                         Comparative Research on Image Segmentation Algorithms”
      In this paper, we have discussed about the effective                                   ,2009 First International Workshop on Education Technology
segmentation algorithm using log edge detector[18],and about                                 and Computer Science.
the segmentation algorithm using canny detector[17],both                              [10] Rafael C.Gonzalez, Richard E.Woods, “Digital
developed by us, for border detection of real skin lesions and                               Image Processing”, second edition, Prentice-Hall, India.
compared their performance in the border detection of real skin                       [11] Rafael C.Gonzalez, Richard E.Woods, Stevan L.Eddins “Digital
lesions. To verify the capability of the segmentation algorithm                              Image Processing using Matlab”
in detecting the border of the lesions for skin lesion diagnosis,                     [12] Wen-Xiong Kang, Qing-Qiang Yang, Run-peng Liang, “The
the algorithm was applied on diversity of clinical skin image                               Comparative Research on Image Segmentation Algorithms”,
containing lesions with noise. The experimental results                                     2009 First International Workshop on Education Technology and
demonstrated the successful border detection of real skin                                   Computer Science.
lesions by the segmentation algorithm using log detector[18]                          [13] Chunming Li, Chenyang Xu , Changfeng Gui, and Martin D. Fox ,
for clinical skin images with noise and make them accessible                               “Level Set Evolution Without Re-initialization: A New Variational
for further analysis and research. We conclude that the                                     Formulation”,Proceedings of the 2005 IEEE Computer Society
segmentation algorithm using log detector[18] segments the                                  Conference on Computer Vision and Pattern Recognition
lesion from the image even in the presence of noise and                                     (CVPR’05)
presence of hair for a variety of lesions, and skin types and we                      [14] J.H.Jaseema Yasmin, M.Mohamed Sathik, S. Zulaikha
conclude that its performance is better than the segmentation                               Beevi ,“Edge Detection Algorithms for Medical Image
algorithm that uses canny detector[17], for border detection of                             Segmentation”, Proceedings of the International Conference on
real skin lesions for noisy skin lesion diagnosis.                                          Intelligent Design and Analysis of Engineering Products
                                                                                            Systems and Computation 9-10 July 2010
                                                                                            (IDAPSC-10),Coimbatore, Pg 63.




                                                                                 35                                   http://sites.google.com/site/ijcsis/
                                                                                                                      ISSN 1947-5500
                                                             (IJCSIS) International Journal of Computer Science and Information Security,
                                                             Vol. 9, No. 5, May 2011


[15] S.Zulaikha Beevi,M.Mohamed Sathik, “A Robust segmentation             [17] J.H.Jaseema Yasmin, M.Mohamed Sathik, S.Zulaikha Beevi,
     Approach for Noisy Medical Images Using Fuzzy Clustering                   “Effective Border Detection of Noisy Real Skin Lesions for Skin
     with Spatial Probability” ,European Journal of Scientific                   Lesion Diagnosis by Robust Segmentation
     Research,Vol. 41,No.3(2010),pp.437-451                                      Algorithm”,International Journal of Advanced
[16] S.Zulaikha Beevi,M.Mohamed Sathik, K.Senthamarai                             Research in Computer Science,Vol. 1, No. 3, Sept-
     Kannan ,J.H.Jaseema Yasmin, “Hybrid Segmentation Approach                   october 2010,pp.110- 116.(Research paper)
     Using FCM and Dominant Intensity Grouping with Region                 [18] J.H.Jaseema Yasmin, M.Mohamed Sathik, S.Zulaikha Beevi,
    Growing on Medical image”International Journal of Advanced                  “Robust Segmentation Algorithm using LOG Edge Detector for
     Research in Computer Science, Vol. 1, No. 2, July-                          Effective Border Detection of Noisy Skin Lesions”, ICCCET 2011,
    August 2010,pp.103-108 (Review Article)                                      pp. 61-66.




                                                                      36                                  http://sites.google.com/site/ijcsis/
                                                                                                          ISSN 1947-5500

				
DOCUMENT INFO
Shared By:
Stats:
views:286
posted:6/5/2011
language:English
pages:7