Data Mining Preprocessing

Document Sample
Data Mining Preprocessing Powered By Docstoc

Innovative methods for extracting skeletons of bone tissues using various preprocessing methods
Dr V. Venkata Krishna1, M. Radhika Mani2, B. V. Ramana Reddy3
 Abstract— The present paper describes various mechanisms of converting a grey level image into binary image. Most of the packages or image processing softwares are capable in generating the binary image from grey level values of pixels. But the problem in these packages or softwares is they do not reveal the method of conversion. And more over various preprocessing measures are required for proper image analysis, understanding etc. to monitor this problem the present paper investigated hoe the skeletons are going to be affected on various preprocessing methods. The present paper presents innovative preprocessing methods which are applied on medical image. A good comparison is made between these methods.

Index Terms—grey level image, binary image, histograms, preprocessing measures, image analysis.

I. INTRODUCTION Thinning and skeletonization have numerous applications in image analysis and computer vision. For several of these applications significant amount of information is lost during the process of binarization. Applying thinning directly to gray scale images is motivated by the desire of directly processing images with gray levels distributed over a range of intensity values. This will avoid shape distortions that may irremediably affect the presence of features in the binary image generated even if an optimal thresholding algorithm is used to produce the binary image. The gray skeleton is a connected subset of a gray scale pattern which consists of a network of lines and arcs centrally placed along local higher intensity regions. Unfortunately, there is no one single agreed upon definition for gray skeletons [1], [2], [4]. Skeletons are classically associated as a medial axis representation that is regenerative (i.e. could be used to generate the object back exactly). Skeletons are not easily digitizable. It is not possible to have a representation that is a medial axis, preserves connectivity, preserves homotopy and exists on the square digital grid. One of these four restrictions

has to be relaxed. This has opened the way to several approximations known as thinning. One can group most of the published algorithms under one of two approaches. The first approach considers the image as a continuous surface in the 3D Euclidean space and use the first and second partial derivatives of this surface to assign the proper topographical label to each pixel [3], [5]. The second approach is based on the repeated application of a removal process that erodes the gray scale pattern until only one pixel thick subset is obtained in the center of the high intensity region. The algorithm proposed in this paper is a parallel thinning algorithm that preserves connectivity and belongs to the latter family. We have devised a set of conditions that guarantee that the resulting thinned version is connected and as close as possible to the medial axis. The algorithm has been tested on a variety of images from different applications and produced satisfactory results that proved to be useful for compression and recognition applications which will be reported elsewhere.

II. METHODOLOGY The present section briefly outlines the various methods of converting grey level image into preprocessed grey levels. The basic structure of this conversion is outlined in figure 1. The preprocessed grey level image is converted into binary by global average.

Figure 1: Block diagram for skeletanization of preprocessed binary image

A binary image can be obtained by various preprocessing methods. The present paper taken into consideration the following preprocessing methods applied on local neighborhoods, which are listed below. local maximum, local minimum, mode, median , mean and ((max-min )/2 ).

Prof. of CSE & Principal CIST, Kakinada, A.P 2 Asst. Prof, Dept. CSE, Member SR Research Forum GIET, Rajahmundry, A.P 3 Member SR Research Forum Research Scholar, JNT University

2 III. RESULTS OF THE PROPOSED APPROACH Fig 3(c)-8(c): Corresponding skeletons of the preprocessed Binary Images IV. CONCLUSION The present paper concludes that there are n numbers of preprocessing methods which can be further expanded. From the figures it is clearly evident that a good skeleton that represents topology is obtained for (max-min)/2 preprocessing method. However good skeletons with out any noise are obtained for median, mean preprocessing methods. But they failed in representing the shape. The final conclusion is “one can not say which preprocessing method is superior; this depends upon the type of image, the way the grey levels are spread, and type of application. There fore 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.

Fig 2

Fig 3(a)

Fig 3(b)

Fig 3(c)

ACKNOWLEDGMENT The authors would like to express their gratitude to Sri K.V.V. Satya Narayana Raju, Chairman, and K. Sashi Kiran Varma, Managing Director, Chaitanya group of Institutions for providing necessary infrastructure. Authors would like to thank anonymous reviewers for their valuable comments and Dr. G.V.S. Ananta Lakshmi for her invaluable suggestions which led to improvise the presentation quality of this paper. REFERENCES
Fig 5(a) Fig 5(b) Fig 5(c) [1] 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, Feb. 1994, pp. 294-299. Arcelli, C. and Ramella, G., ”Finding grey-skeletons by iterated pixel removal”, Image and Vision Computing, vol. 13, no. 3, Apr. 1995, pp. 159-267. 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. Levi, G. and Montanari, U., “A gray-weighted skeleton”, Information and Control, vol. 17, 1970, pp. 62-91. Wang, L. and Pavlidis, T., “Direct gray-scale extraction of features for character recognition”, IEEE Trans. on Pattern Anal. And Machine Intell., vol. 15, no. 10, Oct. 1993, pp. 1053-1067. Samira S. Mersa1, Ahmed M. Darwish, “A New Parallel Thinning Algorithm for Gray Scale Images”, 2002.

Fig 4(a)

Fig 4b)

Fig 4(c)


[3] Fig 6(a) Fig 6(b) Fig 6(c)

[4] [5]


Fig 7(a)

Fig 7(b)

Fig 7(c)

Fig 8(a)

Fig 8(b)

Fig 8(c)

Fig 2: Original Image Fig 3(a)-8(a): Various preprocessed grey level images (Max, Min, (Max-Min)/2, Mean, Median, Mode) Fig 3(b)-8(b): Corresponding Binary Images of preprocessed grey level images

Shared By:
Description: Data Mining Preprocessing