一种AAM模型的MRI医学图像分割算法  

Algorithm of AAM model for MRI medical image segmentation

作　　者：刘金清[1] 吴庆祥[1] 

机构地区：[1]福建师范大学光电与信息工程学院福建省光子技术重点实验室,福州350007

出　　处：《电子测量与仪器学报》2013年第3期236-240,共5页Journal of Electronic Measurement and Instrumentation

基　　金：国家自然科学基金(61179011);福建省自然科学基金(2010J01327)资助项目

摘　　要：提出了一种主动外观模型的MRI医学图像分割的算法。针对主动外观模型中PCA方法对数据细节描述能力的不足,用ICA代替PCA同时进行形状和纹理模型建模,形成基于独立分量分析的改进型主动外观模型。并且在模型中引入了检测到的点与模型点之间的误差的约束项,并提出了平均长度约束项的概念,将它增加到模型中.通过对不同MRI医学图像的实验表明,改进的AAM模型分割的精度和效率都有较大程度的提高。An algorithm of active appearance model for MRI medical image segmentation is proposed. Since PCA in active appearance model lacks for describing data in detail, it is replaced by ICA in both shape and texture modeling, and then an improved ICA-AAMs is proposed. A function, which computes the distance from known position to the corresponding model point, is added in the active appearance model, and then the proposed concept of average length constraint is used in the model. Different MRI medical image experiments show that the algorithm of improved active appearance model segments the image more accurate and efficient than active appearance model.

关 键 词：主动外观模型 主成分分析法 独立分量分析 核磁共振图像 图像分割 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]