一种由内向外的Active Contour模型图像分割算法

针对主动轮廓模型图像分割,提出一种改进算法。该算法首先选取边界内部点为轮廓初始点,再根据能量与面积的关系,将传统算法中初始点向边界点收缩改进为由初始点向边界点膨胀。算法的实现采用原始模型和改进模型相互交替的贪婪算法,保证了模型的稳定性特点。实验结果表明:这种由内向外的分割算法比传统模型分割效果好及适用范围广;避免了传统模型对凹陷区域分割效果不理想及边界外部复杂环境对分割效果的影响。

Image segmentation based on expansive Active Contour model

In the segmentation of region,the contour points are easily affected by concavity boundary and external complex environmental,focusing on these problems,the paper improves energy function of traditional Active Contour model.If initial contour points are put inside the boundary,with relationship between area and energy in the improved model,the contour points will be expanded to boundary,not to be converged in traditional model.The paper uses alternate greedy algorithm of improved model and traditional Active Contour model.Results show that,by the alternate algorithm,the contour points would be expanded quickly and accurately to boundary without disturbance of external environment.