一种改进的活动轮廓的图像分割方法

AN IMAGE SEGMENTATION WITH IMPROVED ACTIVE CONTOUR

对活动轮廓模型存在的问题,即活动曲线收缩范围小且无法检测图像的凹陷边界,提出了一种改进的算法.先采用GREEDY近似最优算法使活动曲线尽可能地接近图像的边界,然后在图像缓变区域取一适当的递减参数对外部能量进行放大,以使活动曲线向图像的凹陷边界移动.实验结果表明该方法是有效可行的.

An algorithm is presented to improve the active contour models, in which have too small capture range and poor convergence to boundary concavities. First, an active curve is made as near as possible to the surfaces by the Greedy Approximate Optimizations Algorithm. And then, a proper decreasing parameter is taken to increase the external energy in the homogeneous regions, which makes the active curve moving to the concave surfaces of image. The experiment showes the method is effective.