A new line symmetry distance based automatic clustering technique: Application to image segmentation

Abstract

AbstractIn this article, at first an automatic clustering technique using the concept of line symmetry property is developed. The proposed real‐coded variable string length genetic clustering technique (VGALS clustering) is able to evolve the number of clusters present in the data set automatically. Here assignment of points to different clusters is done based on the line symmetry based distance rather than the Euclidean distance. The cluster centers are encoded in the chromosomes, whose value may vary. A newly developed line symmetry based cluster validity index, LineSym‐index, is used as a measure of “goodness” of the corresponding partitioning. This validity index is able to correctly indicate the presence of clusters of different sizes as long as they are line symmetrical. A Kd‐tree based data structure is used to reduce the complexity of computing the line symmetry distance. The proposed technique is then applied to automatically segment different images. At first, the superiority of the proposed method to automatically segment the image data sets over Fuzzy C‐means clustering technique, well‐known mean‐shift based method and GAPS clustering with Sym‐index based method, are demonstrated for three remote sensing satellite images. Thereafter it is applied on several simulated T1‐weighted, T2‐weighted, and proton density normal and MS lesion magnetic resonance brain images. The proposed method is able to detect most of the regions well. Superiority of the proposed method over Fuzzy C‐means and Expectation Maximization clustering algorithms are demonstrated quantitatively. The automatic segmentation obtained by VGALS clustering technique is also compared with the available ground truth information. © 2011 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 21, 86–100, 2011