摘要
目的构造水平集方法中所采用的符号距离函数,以提高曲线演化算法的速度。方法通过对水平集曲线形状的分析,给出了曲线内外点的判断方法,然后以距离等于1为步长推进边界,同时进行邻域最近点扫描,寻找任一网格点在曲线上的最近点,从而给出该点到曲线的距离。结果内外点的判断以及符号距离函数的构造均只需要经过图像的一次扫描即可给出结果,计算复杂度为O(N)。结论数值实验分析结果表明,新的方法计算稳定,比其他方法速度快。
Aim To generate the signed distance function of level set.Methods First based on analyzing the shape of level set curve,a new method is presented to define the in-out point of a closed curve.Second,based on analyzing the nearest point of neighbor,the curve can be fraced with a step of distance 1,then the nearest point of one dot could be found among the nearest point of its neighbors,so the distance between one dot of image and curve can be obtained.Results The in-out point of the curve is found and the signed distance function in need for only one time is generated to trace the net point of an image.Conclusion Results of analytic experiments show that this algorithm is robust and faster than the other methods.
出处
《西北大学学报(自然科学版)》
CAS
CSCD
北大核心
2007年第5期697-700,共4页
Journal of Northwest University(Natural Science Edition)
基金
陕西省教育厅专项基金资助项目(JK05303)
关键词
水平集
符号距离函数
邻域
最近点
level set
signed distance function
neighbor
nearest point