一种组合主动轮廓线模型算法

本文针对传统主动轮廓线模型(Snake模型)无法检测凹陷目标轮廓的缺陷，提出了一种由全局Snake模型和局部Snake模型两部分组成的组合Snake模型。组合模型首先使用全局Snake模型进行轮廓粗检测，并使用SUSAN算子检测目标轮廓上凹陷最“深”的凹点；然后，在凹点附近的局部区域，使用局部Snake模型进行轮廓凹陷部分的检测；其后以其替代使用全局模型检测出的目标轮廓的相应部分，形成最终检测的目标轮廓。实验结果表明，本算法具有较好的检测精度和抗噪性。

An Algorithm for the Combined Active Contour Model

The Active Contour Model, or Snake, is used extensively in computer vision and image processing, particularly to locate object boundaries . A combined Snake is presented here to solve the poor convergence to boundary concavities which exists in the traditional Snake. The combined model is made up of two modules: the global model ,which detects the coarse boundary in the early time of the algorithm; and the local model, which can further detect the concave part of the boundary based on the detection of the concave points by using the SUSAN operator. Finally a total fine boundary can be gained with the two models. Experimental results show the algorithm is efficient, precise and immune to the noise.