Abstract: Level set methods have been extensively used in image segmentation，while the traditional level set methods is numerically necessary to keep the evolving level set function close to a signed distance function by periodically re-initializing，re-initializing the level set function is obviously a disagreement between the theory of the level set method and its implementation.Recently，a new variational formulation which completely eliminates the need of the costly re-initialization procedure has proposed by Li C et.at.The main drawback of this model is due to the one direction propagation，i.e.the initial curve either shrinking or expanding towards the object boundaries.In this paper，a new model is proposed subject to binary image for active contours to detect object in a given image，based on techniques of distance preserving level set method.The proposed models is free from the initial condition，furthermore the curve converge to the object boundary precisely，more importantly the curve evolution only take one iteration.

Key words: level set method, distance preserving, geometrical active contour, image segmentation

摘要： 水平集方法已被广泛地应用在图像分割中，传统的水平集方法需要通过周期性的初始化水平集函数使得它一直保持在符号距离函数附近，然而初始化与水平集理论和实现相违背。最近，Li C等人提出一种完全不需要初始化的变分模型，该模型的主要不足就是单方向演化，即演化曲线或收缩或扩张到目标边界。针对二值图像提出一种新的基于距离保持水平集方法的活动轮廓模型，它不依赖于初始位置，演化曲线准确地收敛在目标边界，更重要的是曲线演化只需一次迭代。

关键词: 水平集方法, 距离保持, 几何活动轮廓, 图像分割