基于改进Snake模型的图像分割方法

Image Segmentation Based on Improved Snake Model

Snake(主动轮廓线)模型即能量最小化运动曲线模型,最初由Kass在1987年提出,具有良好的获取特定区域内目标边缘的能力,是一种极为有效的图像分割方法。针对传统Snake模型对初始轮廓的依赖性问题,利用围绕目标形心的圆环间平均灰度差异来确定初始轮廓点,对噪声的干扰有一定的抑制作用,并减少了人工选取的工作量。通过离散Snake算法与分段DP算法的有效结合来获取图像的特征边缘点,以提高Snake算法的收敛速度。最后利用单调性原则对边缘点进行分区,在各个单调区间内采用曲线拟合的方法来获得连续的图像边缘。实验结果表明,基于改进Snake模型的图像分割方法可以从图像中提取连续、封闭的边缘曲线,能够较好的将目标从图像中提取出来。

Snake（Active Contour）Model, introduced by M. Kass in 1987, is a dynamic curve model with energy -minimizing. Snake algorithm, which has advantages in extracting target object from a certain region, is an effective method in image segmentation. Aimed at decreasing the Snake＇s dependence on initial contour points, the present research adopts the difference of mean gray - levels between rings around the object center to obtain the initial contour points, which can reduce artificial selection work as well as noise interference. In addition, the present research tries to combine Discrete Snake Algorithm and Piecewise DP Algorithm to obtain the characteristic edge points of the image so as to improve the convergence speed of Snake. Finally, the research uses monotonic principle to divide those edge points to get some monotone zones, and a curve fitting method is, furthermore, adopted in each monotone zone to get continuous edge of the image. The result shows that to use Improved Snake Algorithm can get a continuous and close edge, thus,the target object can be extracted from the image well.