文章基本信息

标题：An Enhanced Algorithm for Color Image Segmentation
本地全文：下载
作者：Sumant V. Joshi ; Atul. N. Shire
期刊名称：International Journal of Advanced Research In Computer Science and Software Engineering
印刷版ISSN：2277-6451
电子版ISSN：2277-128X
出版年度：2013
卷号：3
期号：4
出版社：S.S. Mishra
摘要：The image segmentation forms the basis for processing and analyzing an image required for most of the applications. For the segmentation of the color images, a novel approach is proposed in this paper by integrating the advantages of the mean shift (MS) segmentation and the normalized cut (Ncut) partitioning met hods. This approach results in effective and robust segmentation of color images and requires low computational complexity and is therefore proved utilitarian for real-time image segmentation processing. The proposed approach is having two phases, preprocessing and post postprocessing. In preprocessing, segmented regions are formed by applying the MS algorithm maintaining the desirable discontinuity characteristics of the image. The graph structure is used to represent the segmented regions and then the Ncut method is applied to perform globally optimized clustering. As conventional graph partitioning methods are directly applied to the image pixels, hence possesses more complexity but in this approach, the complexity is significantly reduces because the num ber of the segmented regions is much smaller than that of the image pixels, therefore allows a low-dimensional image clustering. Operating on regions instead image pixels, also reduces the sensitivity to noise and results in enhanced image segmentation performance. The robustness and effectiveness of the proposed method is tested and verified through a large number of experiments using color images
关键词：Segmentation; Normalized cut; Mean shift Clustering