基于自适应气球力Snake模型的图像分割  

Image segmentation using adaptive balloon force snake model

作　　者：申铉京[1,2] 王开业[1,2] 千庆姬[3] 刘英杰[4] 李想[4] 

机构地区：[1]吉林大学计算机科学与技术学院,长春130012 [2]吉林大学符号计算与知识工程教育部重点实验室,长春130012 [3]吉林大学物理学院,长春130022 [4]中国电信吉林省松原市电信分公司,吉林松原138000

出　　处：《吉林大学学报（工学版）》2011年第5期1394-1400,共7页Journal of Jilin University:Engineering and Technology Edition

基　　金：国家自然科学基金项目(66773098);吉林省自然科学基金项目(201115025);吉林大学研究生创新基金项目(20111063)

摘　　要：针对气球力Snake模型存在对初始轮廓线位置敏感和弱边界泄露的缺点,将图像的全局信息融入到气球力Snake模型中,提出了基于模糊聚类的自适应气球力Snake模型,并应用于图像分割。该模型可以根据图像的区域信息自适应地改变气球力的方向和大小,使得轮廓线能正确收敛到目标边界。实验结果表明:改进模型放宽了对初始轮廓线位置的限制,克服了弱边界泄露的缺点,能使轮廓线在深度凹陷区域得到正确收敛,对噪声的鲁棒性强,且不存在GVF场的临界点问题。The balloon force snake model is sensitive to the location of initial contour and has the shortcoming of weak boundary leakage.An adaptive balloon force snake model based on fuzzy clustering is proposed.This model integrates global information of the image into the balloon force snake model,and is used for image segmentation.It can adaptively change the direction and magnitude of the balloon force according to regional information,which makes the contour converge to the target boundary correctly.Experiment results show that the proposed model has lower restriction on the location of the initial contour and overcome the shortcoming of weak boundary leakage.Therefore,the contour can converge on concave boundary correctly.Furthermore,the improved model has strong robustness to noise and has no critical point problem of Gradient Vector Flow（GVF）.

关 键 词：计算机应用 图像分割 SNAKE模型 自适应气球力 模糊聚类 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]