一种改进的NGVF Snake模型  

An Improved NGVF Snake Model

作　　者：胡学刚[1,2] 汤宏静 

机构地区：[1]重庆邮电大学计算机科学与技术学院,重庆400065 [2]重庆邮电大学系统理论及其应用研究中心,重庆400065

出　　处：《西南大学学报（自然科学版）》2014年第4期139-145,共7页Journal of Southwest University(Natural Science Edition)

基　　金：国家自然科学基金资助项目(11071266);重庆市教委科研基金资助项目(KJ100505)

摘　　要：为解决NGVF Snake模型角点定位精度低和无法收敛到弱边缘的不足,利用GVF场强分布和扩散过程的特性,通过引进两个随空间变化的权重函数,提出了一种改进的NGVF Snake模型.实验结果表明,新模型不仅能保持原GVF Snake模型和NGVF Snake模型捕获区域较大、能很好地收敛到凹陷区域等优点,而且能很好地保护弱边缘和图像细节,提高了图像分割的准确性和分割效率.According to the features of GVF field intensity distribution and diffusion process,a new improved NGVF Snake model is proposed by bringing in two spatially varying weighting functions to solve the problem that NGVF Snake model has low accuracy on locating corner and can erase weak boundaries. The experimental results show that the new Snake model can not only preserve weak edges and image details,but improve the accuracy and efficiency of segmentation while maintaining other properties of GVF and NGVF Snakes such as enlarged capture range and convergence to a U-shaped concavity.

关 键 词：活动轮廓模型 梯度向量流 图像分割 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]