{
"title": "A New Approach for Image Segmentation using Pillar-Kmeans Algorithm",
"authors": "Ali Ridho Barakbah, Yasushi Kiyoki",
"country": null,
"institution": null,
"volume": "35",
"journal": "International Journal of Electrical, Computer, Energetic, Electronic and Communication Engineering",
"pagesStart": 1897,
"pagesEnd": 1903,
"ISSN": "1307-6892",
"URL": "http:\/\/waset.org\/publications\/354",
"abstract": "This paper presents a new approach for image\r\nsegmentation by applying Pillar-Kmeans algorithm. This\r\nsegmentation process includes a new mechanism for clustering the\r\nelements of high-resolution images in order to improve precision and\r\nreduce computation time. The system applies K-means clustering to\r\nthe image segmentation after optimized by Pillar Algorithm. The\r\nPillar algorithm considers the pillars- placement which should be\r\nlocated as far as possible from each other to withstand against the\r\npressure distribution of a roof, as identical to the number of centroids\r\namongst the data distribution. This algorithm is able to optimize the\r\nK-means clustering for image segmentation in aspects of precision\r\nand computation time. It designates the initial centroids- positions\r\nby calculating the accumulated distance metric between each data\r\npoint and all previous centroids, and then selects data points which\r\nhave the maximum distance as new initial centroids. This algorithm\r\ndistributes all initial centroids according to the maximum\r\naccumulated distance metric. This paper evaluates the proposed\r\napproach for image segmentation by comparing with K-means and\r\nGaussian Mixture Model algorithm and involving RGB, HSV, HSL\r\nand CIELAB color spaces. The experimental results clarify the\r\neffectiveness of our approach to improve the segmentation quality in\r\naspects of precision and computational time.",
"references": null,
"publisher": "World Academy of Science, Engineering and Technology",
"index": "International Science Index 35, 2009"
}