FGMM-MRF层次模型在图像分割中的应用

Application of FGMM-MRF Hierarchical Model to Image Segmentation

为了更好地反映图像的区域结构,在高层次标记图像中,区域内部用各向同性的MRF建模,区域的边界用各向异性的MRF来建模;在低层次灰度图像中,用FGMM来描述待分割图像的概率分布.采用Bayes方法,根据标记图像的后验分布所对应的FGMMMRF模型的条件概率,用ICM局部优化算法获得MAP准则下的分割图像.用模拟图像和MR图像进行实验,区域的边界和整体属性具有较好的视觉效果.

In order to accurately describe the region structure of a higher-level label image, the interior region is modeled by isotropic Markov random field （MRF）, while the boundary is modeled by anisotropic MRF. For lower-level gray image, the prior distribution of segmentation image is modeled by finite general mixture model （FGMM）. According to the posterior distribution of the label image conditioned on the gray image corresponding to the conditional probability of FGMM-MRF model, the Bayes formulation and the local iterated conditional modes （ICM） optimization algorithm are adopted, and based on the MAP （maximum a posterior） criterion the image segmentation result is obtained. Numerical simulations demonstrate that the whole property and the boundary of image area show better vision effect with a test to synthetic image and real MR brain image.