基于改进的参数活动轮廓模型的图像分割方法

Image segmentation based on improved parametric active contour model

参数活动轮廓模型(SNAKE模型)被广泛应用于医学图像分割中。传统的SNAKE模型对初始轮廓非常敏感,必须给定在图像边缘附近,否则容易陷入局部极小而无法收敛到真实轮廓处。提出一种基于距离变换的距离势能模型作为自适应外力,能很好地引导SNAKE逼近真实边界而不依赖于特定的初始轮廓的位置。实验结果表明:该算法快速有效,能在更大的范围内捕捉图像特征,是一种有效的图像分割的算法。

The parametric active contour model is widely used in the medical image segmentation. The original model is very susceptible and depends on the initial contour which must be set near the edge of the image, or the SNAKE points are easily trapped by the local minimum values. A kind of the potential energy based on the distance transformation as the automatic external force is presented, which can lead the active contour to find the real edge without any special initialization of the contour. The experiment results show that the method is characterized by its rapidity and capacity which can capture the image features in a wider region and is an effective algorithm to segment the medical image.