Abstract: The level set based Chan-Vese model is a representative geometric active contour model that has been successfully applied to a variety of image segmentation problems. To enhance the computational efficiency and the segmentation accuracy of the model, an efficient numerical algorithm is developed to solve the Chan-Vese model by using the radial basis point interpolation. By approximating the level set function with the radial basis point interpolation, the Chan-Vese model is discretized into an initial?value?problem?for?ordinary?differential?equations, which can be solved by the forward Euler’s method. This algorithm is free of grids and insensitive to initial contours, gets rid of the time-consuming re-initialization, and has a stop criterion with no presetting iteration number. Experimental results show that the algorithm segments images accurately even with no initial contour and possesses high segmentation speed.

Key words: image segmentation, Chan-Vese model, radial basis function, radial basis point interpolation, level set

摘要： 基于水平集方法的Chan-Vese模型是一种典型的几何活动轮廓模型，已成功应用于众多领域中的图像分割问题。为了提高该模型的演化速度和分割效果，提出了一种基于径向基点插值求解Chan-Vese模型的高效数值算法。通过用径向基点插值法逼近水平集函数，Chan-Vese模型被离散为常微分方程组初值问题并可用向前Euler法求解。该算法不需要网格单元，对水平集初始轮廓不敏感，不涉及复杂费时的重新初始化过程，并且有明确的演化终止条件，无需事先设置演化次数。实验表明该算法在没有初始轮廓时也能正确分割图像，具有很快的演化速度。

关键词: 图像分割, Chan-Vese模型, 径向基函数, 径向基点插值法, 水平集