模糊聚类协作区域主动轮廓模型医学图像分割  

Active contour model-based medical image segmentation method collaborative with fuzzy C-means

作　　者：李宣平[1] 王雪[1] 

机构地区：[1]清华大学精密仪器与机械学系精密测试技术及仪器国家重点实验室,北京100084

出　　处：《仪器仪表学报》2013年第4期860-865,共6页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金项目(61272428);教育部博士点基金项目(20120002110067)资助

摘　　要：医学图像分割的研究对于医学影像发展具有重要意义。区域主动轮廓模型(CV)易受目标和背景区域面积比的影响,且对初始位置敏感。针对上述现象,本文提出一种模糊C-均值聚类(FCM)协作改进CV模型的图像分割算法,即FCM-CV算法。首先在CV模型中增加能量权值函数消除面积比的影响,然后用FCM粗分割结果指导设定改进CV模型零水平集的初始位置。实验结果表明,与CV模型和局部二值拟合模型(LBF)相比,FCM-CV算法消除了面积比对分割精度和效率的影响,具有更好的数值稳定性,且对初始位置不敏感,提高了图像分割的准确性。The study of medical image segmentation is of great significance to the development of medical imaging. The region active contour model（ CV model） is subject to the influence of the area ratio of the target and background regions of the image to be segmented, and is sensitive to the initial contour of zero level set. Aiming at these prob- lems,this paper proposes an image segmentation method, i. e. FCM-CV collaborative image segmentation algorithm based on FCM and CV model. First, energy weight functions are added to the CV model in order to eliminate the effect of the area ratio on the model. Then the coarse segmentation result of FCM is applied to set the initial contour of zero level set of the improved CV model. Experiment results indicate that compared with conventional CV model and LBF model, the proposed FCM-CV algorithm eliminates the effect of area ratio on segmentation precision and effi- ciency,and has better numerical stability. Furthermore,the FCM-CV algorithm is insensitive to initial contour of zero level set and improves the accuracy of image segmentation.

关 键 词：图像分割 CV模型 能量权值函数 模糊C均值聚类 协作 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]