一种结合位置信息的混合模型图像分割方法  

A Mixture Model with Spatial Constraint for Image Segmentation

作　　者：于林森[1] 张田文[1] 

机构地区：[1]哈尔滨工业大学计算机科学与技术学院,哈尔滨150001

出　　处：《信号处理》2007年第3期411-414,共4页Journal of Signal Processing

摘　　要：提出了一种新的受位置约束的混合模型图像分割方法。该方法在独立混合模型的基础上,采用空间滤波方法对像素所属分量的后验概率进行修正,在混合模型中隐含地加入了像素的空间位置信息。这种结合位置信息的方法为混合模型的分量选择提供了一种有效的实现方式。与其它的受位置限制的混合模型相比,该方法没有引入额外的模型参数,并且无需采用模型选择准则,可以实现自动的混合分量个数的选择。The paper presents a spatially constrained mixture model for image segmentation. Based on the EM algorithm for independent mixture model, a spatial filtering process is introduced to correct the independent assumption of the pixels. The advantage of the proposed algorithm is that it provides an effective implementation of model selection. Compared with other spatially constrained mixture models,the proposed algorithm is simple and do not need a model selection criterion to choose the suitable number of mixture components. Experiments on synthetic and real images are presented to show the efficacy of the proposed algorithm.

关 键 词：图像分割 混合模型 模型选择 滤波 

分 类 号：TN929.53[电子电信—通信与信息系统]