基于跟踪算法的肺部CT图像血管提取

在肺部CT图像中,血管与背景的对比度较低,因此传统的阈值细化方法只能提取有限的一部分血管。为了解决这个问题,我们在阈值细化方法处理后,以检测出的血管为基础,利用血管与背景梯度信息跟踪检测出在阈值细化方法中漏检的血管。采用OTSU算法二值化肺部CT图像,形态学细化算法提取血管骨架,依据血管光滑性确定跟踪方向并根据灰度梯度信息跟踪出更多的细小血管。实验说明该方法是有效的。

Extraction of blood vessel in lung CT images based on tracking algorithm

Because of the low contrast between blood vessel and background in lung CT images,a part of vessel can be extracted through the traditional thresholding and thinning algorithm.To solve this problem,on the basis of the vessel detected by thresholding and thinning algorithm,a tracking algorithm depending on the gray gradient between vessel and background is proposed in the paper,which extracts the vessel missed by thresholding and thinning processing.In this paper,we extract the skeleton of vessel by using morphological thinning algorithm from OTSU algorithm based binarization image,and more tiny vessel tracked by using gradient information and direction depending on the smoothness of vessel.It is proved effective by experiment.