摘要
介绍了一种改进的live wire和轮廓插值技术相结合的医学图像序列分割方法。在这个方法中,先运用分数阶微分算子代替传统的Sobel、Canny等算子构造新的代价函数,对live wire算法进行改进,并用改进的live wire算法对医学图像序列中的关键层切片进行交互式的准确分割;然后根据这些关键层切片的分割结果,采用非线性的轮廓插值技术对中间层切片进行自动分割;最后,对分割后的医学图片序列进行交互式修正。实验表明该文提出的方法仅需少量的人机交互就能从医学图像序列中分割出感兴趣的物体。
In this paper, we propose an algorithm for the segmentation of medical image series based on the combination of the modified live wire algorithm and the contour interpolation algorithm.At the first, we modify the live wire algorithm by using fractional differential to construct the cost function., and then we do interactive segmentation of the key layers of medical image series;According to the consequence of segmentation, we segment the other images automatically by nonlinear contour interpolation algorithm; Finally, doing some reparation of the automatic segmentation result to guarantee the reliability of the result in interactive way, Experiment shows that this algorithm can obtain the boundarY of the desired object from a Series of medical image quickly and reliably with only little user intervention.
出处
《电子质量》
2009年第11期1-4,共4页
Electronics Quality
关键词
医学图像分割
分数阶微分
LIVE
WIRE
插值
medical image segmentation
fractional differential
live wire
contour interpolation