改进Snake模型在医学超声图像分割中的应用

因其本身所具有的复杂性,医学超声图像的分割足一件非常困难的事情.Snake算法能够有效地利用图像局部与整体的信息,实现对边界的准确定位,因此非常适用于医学超声图像的处理.针对传统Snake模型对初始轮廓依赖性的问题,根据图像边缘附近灰度差异显著的特性,提出了获取初始轮廓点的方法--扇形法,运用方法可以获得靠近日标边界的初始轮廓点,并且得到的轮廓点是有序点集.能量最小化过程运用贪婪算法来获得图像的特征边缘点,最后采用分段多项式拟合的方法来获得连续的图像边缘.实验结果表明.所采用的方法可以获得连续、封闭的边缘曲线,能够较好的将目标从图像中提取出来.

Medical Ultrasound Image Segmentation based on Improved Snake Model

Due to its complexity,the medical ultrasound image segmentation is very difficult.Snake algorithm can utilize image's local and whole information effectively to realize boundary's accurate localization, thus it is suitable to medical ultrasound image processing.Aimed at decreasing the Snake' s dependence on initial contour,this paper puts forward a method fan-shaped method,which rests on the remarkable difference of gray-level nearby the object edge,for the initial contour point selection.By using this meth...