一种改进的Snake模型图像分割算法

The image segmentation of the complex region based on snake model

针对Snake模型不能收敛到深凹陷区域的缺点,提出一种Snake模型的改进算法。首先把模型不能收敛到深凹陷区域归结为内部能量产生的收缩不足,在此基础上改变新的力,以调整内部能量;然后在不增加参数个数的前提下给出新的能量表达式。算法的实现采用贪婪算法。结果表明,改进的Snake模型能迅速地收敛到凹陷区域,并减少了结果对初值的依赖,明显优于传统Snake模型。

To resolve the problems that the traditional snake model can not converge the concavity district well in the segmentation of complex region, the paper makes improvement for the traditional snake model in two aspects. On one hand, increasing one energy item created by doing work, on another hand, adopting greey algorithm in two stages. The result shows that the improved snake model can quickly converge the concavity region and the more complex region, and has decreased the dependence of the result of segmentation to the initial value and the influence of noise point to the boundary.