摘要
现今的停止项采用的全局阈值法很难使经典水平集算法得到较准确的分割效果。为此,构建了一个新的停止项函数,该函数在梯度较小时为一个常数,超过某个阈值后为单调递减函数,直至到某个梯度时递减为零。这保证了采用该函数的水平集方法能够实现在同质区加速演化,在边缘区停止演化的目的。实验结果表明,采用新的停止项函数能够使水平集方法获得更好的分割结果。
It is difficult to get accurate segmentation results of classical level set when its stop function uses the single globe threshold. So a new stop function of level set was proposed. The new stop function would be a constant when the gradient is low. And after the gradient exceeding a certain threshold value, the new stop function became a monotone descending function until reduced to zero at a certain gradient. That could make the level set evolving faster in homogeneous area and stop in edge. Finally, the experimental results show that using the developed stop function based level set is verified well than the traditional one.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2008年第22期6154-6157,共4页
Journal of System Simulation
基金
航空科学基金(20070153005)
航空支撑基金(07C53007)
高等学校博士学科专项科研基金(20060699024).
关键词
水平集
停止项函数
图像分割
局部阈值
level set,stop function,image segmentation,local threshold
