A parallel thinning algorithm for contour extraction and medial axis transform Gomathi Kasi Viswanathan Centre for Information Technology and Engineering, Manonmaniam Sundaranar University, Tirunelveli, India Gomathikv.1991@gmail.com Anita Murugesan Centre for Information Technology and Engineering, Manonmaniam Sundaranar University Tirunelveli, India anitamtech91@gmail.com Krishnan Nallaperumal Professor and Head Centre for Information Technology and Engineering,Manonmaniam Sundaranar University, Tirunelveli, India Krishnan@computer.org Abstract— The thinning algorithm proposed in this paper, is an improved parallel thinning algorithm that aims in both, medial axis based thinning and contour extraction in a same process. In the process of thinning to obtain the medial axis of the character the points are deleted from outer boundary first and then proceeded inside till a single pixel wide skeleton is produced. In this process the outermost boundary line alone is preserved which is then produced as the contour of the image. So within one algorithm we get the two major categories of skeletonization, Medial axis Transform and contour extraction. Experiments are done on printed English characters and the results show that this algorithm is very effective. Keywords— medial axis transform, contour extraction, parallel thinning algorithm, connectivity. I. INTRODUCTION Thinning is defined as the process of reducing generally elongated patterns to a line-like representation (Lam et al., 1992). The output from the thinning process is called a skeleton. Thinning is most widely applied in optical character recognition. Optical character recognition (OCR) is the process of converting scanned images of machine printed or handwritten text into a computer processable format. Generally thinning algorithms are applied on binarised images only. Thinning when applied to a binary image, produces another binary image as output [2]. The output of thinning process, a line drawing representation of a pattern is called a ‘thinned image’. The term ‘skeleton’ can also be used in general to denote a representation of a pattern by a collection of thin arcs and curves. In recent years, thinning and skeletonization have become synonyms in the literature, and the term ‘skeleton’ is used to refer to the result, regardless the shape of the original pattern or the method employed[6].Thinning process is commonly used for two purposes – medial axis transform and contour extraction . Contour Extraction: Contour extraction is the process of extracting the outer boundary lines of the image. Medial axis Transform: Medial axis is the process of extracting central line of the image. Need for Thinning: Need for thinning of images has the following reasons: i) It reduces the amount of data to be processed, as a result time required for processing is reduced. ii) Topology is preserved. iii) One pixel wide skeletons produced are very useful for the purpose of pattern recognition when they use vectorization algorithms. iv) Shape analysis is made easy. Characteristics to be preserved by Skeletons[4]: The skeletons must preserve the following characteristics: a)Geometrical charateristics the skeleton must be in the middle of the original object and must be invariant to translation, rotation, and scale change. b) Topological characteristics: The skeleton must retain the topology of the original object. II. SOME DEFINITIONS: A. Image Binarisation: The image given as input is a black and white pixels. After binarisation the black pixels are represented with 1s and white pixels are represented with 0s. B. 8-Pixel neighborhood and 4-pixel neighborhood: Any pixel X in the image has 8- pixels(X1,X2,X3,X4,X5,X6,X7,X8) surrounding it. Considering all the pixels is known as 8-pixel neighborhood whereas considering only the four pixels (X2,X4,X6,X8) is known as 4-pixel neighborhood. For this algorithm 8-pixel connectivity is considered. 2013 IEEE International Conference on Emerging Trends in Computing, Communication and Nanotechnology (ICECCN 2013) 978-1-4673-5036-5/13/$31.00 © 2013 IEEE 606