A New Modified Gaussian Mixture Model for Color-Texture Segmentation | Science Publications

Journal of Computer Science

M. Sujaritha¹  S. Annadurai¹ 

关键词: Gaussian mixture model;    Expectation Maximization (EM);    Bayesian pixel classification;    color texture segmentation;    MAP estimation;    EM algorithm;    Maximum A Posteriori (MAP);    Spatially Variant Finite Mixture Model (SVFMM);    Markov Random Field (MRF);    Probabilistic Rand Index (PRI);    Boundary Displacement Errors (BDE);    Variation of Information (VoI);   

DOI  :  10.3844/jcssp.2011.279.283

学科分类：计算机科学（综合）

来源: Science Publications

【 摘 要 】

Problem statement: This study presents a new, simple and efficient modified Gaussianmixture model based clustering algorithm for color-texture segmentation. The proposed mixture modelintroduces a new component density function which incorporates spatial information and the weightingfactor for neighborhood effect is fully adaptive to the image content. Approach: It enhances thesmoothness towards piecewise-homogeneous segmentation and reduces the edge-blurring effect. AnExpectation Maximization (EM) model fitting Maximum A Posteriori (MAP) algorithm segments theimage by utilizing the pixel

【 授权许可】

Unknown