摘要
在分析心脏MR图像特点的基础上,提出了先对心脏MRI图像进行K均值聚类,把K均值聚类后的图像作为特征图像,在特征上用Song和Chan提出的快速分割方法进行粗分割,再用粗分割的曲线作为水平集的初始曲线,在心脏MRI图像上用Chan和Vese方法进行细分割的心脏MR图像分割方法。并对Song和Chan快速算法中扫描图像的区域进行了改进,提高了分割速度。分割实验证明,用该方法能够快速、准确地分割心脏MRI图像。
Based on the analysis of the characteristics of cardiac MRI image, a fast segmentation algorithm for cardiac MRI image is proposed. Firstly,the cardiac MRI image is processed by K-means clustering, and the image processed is taken as the new feature for rough seg- mentation by Song and Chan method. Secondly,the curve of rough segmentation acts as the initiative curve of level sets, and Chan and Vese method is applied to the delicate segmentation of cardiac MRI image. In order to improve the segmentation speed, Song and Chan method is improved by reducing the scanning area of image. The experiment shows that the cardiac MRI image can be segmented fast and accurately by the approach proposed.
出处
《计算机应用与软件》
CSCD
北大核心
2007年第11期37-40,共4页
Computer Applications and Software
基金
香港特区政府研究资助局资助项目(CUHK/4180/01E
CUHK/1/00C)
