期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2017
卷号：95
期号：14
页码：3365
出版社：Journal of Theoretical and Applied
摘要：Image segmentation is the most common method used to analyze and locate deformities in medical images. Clustering is a technique used in segmentation to group up similar data in a single cluster. FCM is the successful clustering technique employed, which is repeatedly modified by the researcher to make it more robust against noise. In this paper, we present an improved kernel FCM algorithm for image segmentation by introducing a regularization parameter with covariance weighted fuzzy factor and a Gaussian kernel metric. It calculates and uses the heterogeneity of gray scales in the neighborhood location for acquiring local contextual information and replaces the standard Euclidean distance with Gaussian radial basis kernel functions. The main advantages are adaptiveness to local context, preserve image details with enhanced robustness, independence of clustering parameters and decreased computational costs. The experiments are done in Brain Magnetic Resonance Images (MRI) and results are analyzed, show that the proposed algorithm is better segmentation accuracy and efficiency than the other methods.
关键词：Fuzzy c-means; Gaussian Kernel; Medical images; Weighted Covariance; Magnetic Resonance Imaging
