一种基于区域相关性的三维核磁共振图像快速分割方法  

A Fast Approach to the Segmentation of 3D MRI Images Based on Region Relativity

作　　者：刘兆[1] 沈辉[1] 高政[1] 胡德文[1] 

机构地区：[1]国防科技大学机电工程与自动化学院,湖南长沙410073

出　　处：《计算机工程与科学》2008年第7期54-56,76,共4页Computer Engineering & Science

基　　金：国家973计划资助项目(2003CB716104);国家自然科学基金资助项目(60771062)

摘　　要：本文针对三维核磁共振图像的自动分割问题,结合分水岭算法与SharonE等人提出的SWA分割算法,提出一种基于区域相关性的快速分割算法。该算法的优点是减少了计算复杂度,计算速度得到很大提高,能准确有效地提取出感兴趣的目标区域。实验结果表明,该算法能对一般的三维核磁共振图像及具有复杂纹理的单张切片图像进行准确的分割。This paper gives an automatic segmentation method of 3D MRI images. A fast segmentation algorithm based on region relativity is proposed, combining with the watershed algorithm and the SWA method proposed by Eitan Sharon. The obvious advantage of the algorithm is that the computational cost is reduced and the computational speed is improved greatly. The interested object can be extracted rapidly and effectively. The experimental results show that the new algorithm can accurately segment 3D MR images and single images with a complex texture.

关 键 词：图像分割 图论 分水岭 核磁共振成像 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]