International Journal of Application or Innovation in Engineering & Management (IJAIEM)
Web Site: www.ijaiem.org Email: editor@ijaiem.org, editorijaiem@gmail.com
Volume 2, Issue 6, June 2013 ISSN 2319 - 4847
Professor, Dept. of CSE, M.G.I.T, JNTUH, Hyderabad, Andhra Pradesh, India
A BSTRACT
The present paper describes novel and effective methods of shape representation by combining innovative preprocessing techniques and morphological skeletonization method. Preprocessing mechanism describes various methods of converting a grey level image into binary image. In this paper, six different preprocessing techniques are studied. They are mean, median, mode, maximum, minimum and range (i.e max-min). The binary images obtained from those preprocessing techniques are then subjected to skeletonization using Medial Axis Transform (MAT). Skeletonization is a transformation of a component of a digital image into a subset of the original component. In other words, the most possible reduced image (skeleton) is obtained.
In this paper, shape features are evaluated based on innovative preprocessing techniques and MAT method. The skeletons extracted from each of the binary images obtained from the six techniques are compared with each other. Thus, it is found that all preprocessing techniques provide us with more clear and contiguous skeletons except the range method. The max-min (i.e range) technique helps in extracting the edge of any given binary image. The novelty of this paper is that, it identifies the preprocessing technique which extracts edge for any given binary image instead of skeleton. The experimental results on various shape images clearly demonstrate the efficiency of the proposed method. This method has been tested on different images and the results proved the quality of the approach.
Keywords: Preprocessing, Skeletonization, Edge, Medial Axis Transform.
1.
I NTRODUCTION
Shape representation is a key technology in computer vision and a hot spot in many fields, such as image processing, artificial intelligence etc. In digital image processing skeletonization has been around for many years[1, 2]. Skeleton is an important characteristic of an object and has special advantage in representing the main shape of an object. It is used in image recognition to describe an object and now has become one focus of the research. Skeletonization is the process of reducing foreground regions in a binary image to a lower dimensional subset preserving the original foreground connectivity. It is normally only applied to binary images, and produces another binary image as output. The term
‘skeleton’ has been used in general to denote a representation of a pattern by a collection of thin arcs and curves. Other nomenclatures have been used in different context. For example the term ‘medial axis’ is used to denote the locus of centers of maximal blocks. Some authors also refer to a ‘skeletonized image’ as a line drawing representation of pattern. In recent years, it appears that skeletonization has become synonymous in the literature, and the term
‘skeleton’ is used to refer to the result, regardless of the shape of the original pattern or the method employed. Thus, skeletonization is defined as process of reducing the width of pattern to just a single pixel. There are various applications using this type of image processing in almost every technical field, especially in medical sciences. Other areas like computer vision, image analysis etc., also use this technique. There are several hundreds of algorithms to obtain skeletonization but, the objective is to obtain more clear and contiguous skeleton out of any binary image. For several of these applications significant amount of information is lost during the process of binarization[3], [4], [5].
Preprocessing is one of the methods that help to extract desired skeleton using several techniques [6].
The present paper is organized as follows. The section 2 describes the methodology, the results and discussions are given in section 3 and conclusions are listed in section 4.
2.
M ETHODOLOGY
2.1
Preprocessing Techniques
The present section briefly outlines the various methods of converting grey level image into preprocessed image. The basic structure of this conversion is outlined in figure 1. A binary image can be obtained by various preprocessing methods. The present paper taken into consideration the following preprocessing methods applied on local
Volume 2, Issue 6, June 2013 Page 273

International Journal of Application or Innovation in Engineering & Management (IJAIEM)
Web Site: www.ijaiem.org Email: editor@ijaiem.org, editorijaiem@gmail.com
Volume 2, Issue 6, June 2013 ISSN 2319 - 4847 neighborhoods, which are listed as follows (a) local maximum, (b) local minimum, (c)mode (d)median (e)mean and
(f)max-min (i.e. range). Primarily these preprocessing techniques are adopted to convert gray scale images into their respective binary images. This is performed using any one of the above given methods. On obtaining the binary image, we further reduce it to its corresponding skeleton using the method of Medial Axis Transform. The obtained skeletons are analyzed to determine which preprocessing methods leads in extraction of edge to the given image. This is practically illustrated in the following block diagram given in Figure 1.
Apply different preprocessing methods
Input
Image
Obtain a binary image for each preprocessing technique
Apply skeletonization method on each binary image
Identify the preprocessing method which extracts edge
Figure 1 Block diagram for edge extraction
2.2
Skeleton and shape
Skeletonization is a global space domain technique for shape representation [7]. It has been studied extensively since skeletons have attractive properties which make them suitable for structural pattern recognition [7, 8]. There are two types of skeletonization methods: pixel-based and non-pixel-based. In a pixel-based method, all pixels inside a shape are used in the skeletonization process. Pixel based methods often use thinning techniques [8, 9] or distance transforms
[9,10]. In a non-pixel-based method, only the contour pixels of a shape are used for skeletonization. The skeleton of the shape is analytically derived from its contour [11, 12]. Skeletonization has been a part of image processing for a wide variety of applications [8]. Using this proposed method both contour pixels and pixels inside any shape can be extracted for shape representation.
2.2.1
Medial Axis Transform
The informal definition of a skeleton is a line representation of an object that is: i) one-pixel thick ii) through the middle of the object and iii) preserves the topology of the object. These are not always realizable. Fig. 2 shows counter examples for skeleton definition.
(a) (b)
Figure 2 Counter examples to the three Requirements.
In the first example, Figure 2 (a), it is not possible to generate a line that is one pixel thick and in the center of an object while generating a path that reflects the simplicity of the object. In Figure 2 (b) it is not possible to remove a pixel from the 8-connected object and simultaneously preserve the topology, the notion of connectedness of the object.
Nevertheless, there are a variety of techniques that attempt to achieve this goal and to produce a skeleton.
The MAT is unique. The original image can be reconstructed from its medial axis transform. The MAT evolved through Blum’s work on animal vision systems [13, 14, 15, 16, 17, 18, 19, 20, 21, 22]. His interest involved how animal vision systems extract geometric shape information. There exist wide range applications that find a minimal representation of an image useful [19]. The MAT is especially useful for image compression since reconstruction of an image from its MAT is possible. A basic formulation is based on the work of Lantuejoul. The skeleton subset Sk(A) is defined by the following equation 1.
Skeleton subsets – S k
( A )  E ( A , kB )  [ E ( A , kB )  B ] (1)
Where k=0,1,…….K
Volume 2, Issue 6, June 2013 Page 274

International Journal of Application or Innovation in Engineering & Management (IJAIEM)
Web Site: www.ijaiem.org Email: editor@ijaiem.org, editorijaiem@gmail.com
Volume 2, Issue 6, June 2013 ISSN 2319 - 4847 where K is the largest value of k before the set S k
(A) becomes empty. The structuring element B is chosen (in Z
2
) to approximate a circular disc, that is, convex, bounded and symmetric. The skeleton is then the union of the skeleton subsets as given by the equation 2.
Skeleton – S ( A )  k
K

S
 0 k
( A ) (2)
The original object can be reconstructed with the given knowledge of the skeleton subsets S k
(A), the structuring element B, and k as given by the following equation 3.
Reconstruction – A 
K
 k  0
( S k
( A )  kB ) (3)
However, the skeleton computed by the equation 2 does not preserve the topology and are often sensitive to noise and geometrical transformations, such as rotation and scaling, in particular for high resolution images. This is a common problem for most skeletonization methods [8]. The present study is aimed to overcome the above. For this the present paper proposes a novel approach by combining preprocessing techniques and skeletonization based on Medial Axis
Transform approach. One of the disadvantages of the existing skeletonization method is that it is not automatic. It always requires human interaction for selecting the number of iterations to detect the good skeleton.
3.
R ESULTS AND D ISCUSSIONS
In this paper a given image is converted into binary using various preprocessing methods. To each obtained binary image Medial Axis Transform is applied that is regenerative (i.e. could be used to generate the object back exactly). In this paper the MAT Algorithm is applied on various images on subjection of this image to various preprocessing techniques: maximum, minimum, mean, median, mode and range respectively and the results are as shown in Fig. 3 to
Fig 14. A good comparison is made between these methods. Finally, the results obtained are closely analyzed to identify an effective preprocessing method that gives the edge.
(a) (b) (c)
Figure 3
(a) (b) (c)
Figure 4
Volume 2, Issue 6, June 2013 Page 275

International Journal of Application or Innovation in Engineering & Management (IJAIEM)
Web Site: www.ijaiem.org Email: editor@ijaiem.org, editorijaiem@gmail.com
Volume 2, Issue 6, June 2013 ISSN 2319 - 4847
(a) (b) (c)
Figure 5
(a) (b) (c)
Figure 6
(a) (b) (c)
Figure 7
(a) (b) (c)
Figure 8
Figures 3 to 8 represent edge extraction method in which Figures 3(a) to 8(a) represent original images, Figures 3(b) to
8(b) represent corresponding skeleton of preprocessed binary images and Figures 3(c) to 8(c) are reconstructed images.
Volume 2, Issue 6, June 2013 Page 276

International Journal of Application or Innovation in Engineering & Management (IJAIEM)
Web Site: www.ijaiem.org Email: editor@ijaiem.org, editorijaiem@gmail.com
Volume 2, Issue 6, June 2013 ISSN 2319 - 4847
(a) (b) (c)
Figure 9
(a) (b) (c)
Figure 10
(a) (b) (c)
Figure 11
(a) (b) (c)
Figure 12
(a) (b) (c)
Figure 13
Volume 2, Issue 6, June 2013 Page 277

International Journal of Application or Innovation in Engineering & Management (IJAIEM)
Web Site: www.ijaiem.org Email: editor@ijaiem.org, editorijaiem@gmail.com
Volume 2, Issue 6, June 2013 ISSN 2319 - 4847
(a) (b) (c)
Figure 14
Figures 9 to 14 represent edge extraction method in which Figures 9(a) to 14(a) represent original images, Figures 9(b) to 14(b) represent corresponding skeleton of preprocessed binary images and Figures 9(c) to 14(c) are reconstructed images.
4.
CONCLUSIONS
The present paper proposes an innovative approach for the extraction of edge based on preprocessing techniques and
Medial Axis Transform scheme, which is completely based on morphology, for the recognition purpose. It is found that two types of shape representation (edge and skeleton) can be obtained by applying various methods in preprocessing on
Medial Axis Transform approach with any given binary images.
The present paper concludes that there are n numbers of preprocessing methods which can be further expanded. From the results it is clearly evident that a good edge is obtained for (max-min)/2 preprocessing method. The final conclusion is “one cannot say which preprocessing method is superior; this depends upon the type of image, the way the grey levels are spread, and type of application. Therefore the present paper recommends it is better to choose one of the preprocessing methods by applying and comparing each instead depending upon a constant method.
References
[1] Jain, A.K., Fundamentals of Digital Image Processing, Prentice-Hall, 1986.
[2] Gonzalez, R. and Woods, R.E., Digital Image Processing , Prentice Hall, 2002.
[3] Abe, K., Mizutani, F. and Wang, C., “Thinning of grayscale images with combined sequential and parallel
conditions for pixel removal”, IEEE Trans. on Systems Man Cybernetics, vol. 24, no. 2, pp. 294-299, Feb. 1994.
[4] Arcelli, C. and Ramella, G.,”Finding grey-skeletons by iterated pixel removal”, Image and Vision Computing, vol.
13, no. 3, pp. 159-267, Apr. 1995.
[5] Levi, G. and Montanari, U., “A gray-weighted skeleton”, Information and Control, vol. 17, pp. 62-91, 1970.
[6] Abdulla, W.H., Saleh, A.O.N. and Morad, A.H., : A preprocessing algorithm for hand-written character recognition, pattern recognition, letters 7,1998
[7] Loncaric S., “A Survey of Shape Analysis Techniques,” Pattern Recognition, Vol. 31, pp. 983 – 1001, 1998.
[8] Lam L., Lee S., and Suen C., “Thinning methodologies –A comprehensive survey,” IEEE Transactions on Pattern
Analysis and Machine Intelligence, Vol. 14, pp. 869-885, June 1992.
[9] Smith R.W., Computer Processing of Line Images: A Survey,” Pattern Recognition, Vol. 20, pp. 7–15, 1987.
[10] Chang H.S. and Yan H., “Analysis of Stroke Structures of Handwritten Chinese Characters,” IEEE Trans. Sys. (B)
29 (1999) 47–61.
[11] Man Cybern B. et al., spline contour representation and symmetry detection,” IEEE Trans. Pattern Anal. Mach.
Intell., Vol. 15, pp. 1191–1197, 1993.
[12] Zou J.J. and Yan H., “Skeletonization of Ribbon-Like Shapes Based on Regularity and Singularity Analyses,”
IEEE Trans. Sys. Man Cybern. (B), Vol. 31, pp. 391–395, 2001.
[13] Chen, S. and Shih, F., “Skeletonization for fuzzy degraded character images”, IEEE Trans. on Image Processing,
vol. 5, no. 10, Oct. 1996, pp. 1481-1485.
AUTHOR
Dr. M. Rama Bai received, her B.E degree from Bharathiar University, Coimbatore(T.N) and her M.Tech
(CSE) from College of Engineering, Osmania University, Hyderabad. She received her Ph.D. degree in
Computer Science from Jawaharlal Nehru Technological University, Kakinada (JNTUK) in 2012. She served Amrita University, Coimbatore and Sri Hindu College of Engineering, Machilipatnam before joining in MGIT for some period. She then joined as Assistant Professor in the Dept of Computer Science & Engineering,
Mahatma Gandhi Institute of Technology (MGIT) in 1999. At present she is working as Professor in Dept of CSE,
MGIT. Her research interests include Image Processing, Pattern Recognition, Digital Water Marking and Image
Retrieval Systems. She has published 18 research publications in various National and International Journals and conferences.
Volume 2, Issue 6, June 2013 Page 278