基于改进区域生长算法的肝脏管道图像分割方法

摘要

In the process of image segmentation, the selection of seed points of traditional region growing method requires artiifcial judgement, which will increase the workload and decrease the work efifciency. In order to reduce the amount of user interaction in the selection of seed points, this paper puts forward an improved region growing algorithm based on the prediction of the seed point locations. According to the representative characteristics of vascular skeleton lines, the location of hepatic ducts in the image can be predicted by the mapping transformation of the coordinate, which will reduce the manual participation and ascertain the number and locations of seed points in the image automatically. The ideal segmentation results of a liver image indicate that the algorithm can accomplish the prediction of seed points with less user interaction.%在图像分割过程中，传统区域生长法种子点的选取需人工判定，工作量较大，效率较低。为了减少种子点选取时的用户交互量，本文提出了一种基于种子点位置预判的改进区域生长算法。该算法基于血管骨架线具有代表性的特点，通过坐标系的映射转换来预判图像中肝脏管道的位置，减少了人工参与，实现了图像种子点数目和位置的自动确定。肝脏图像分割实验结果表明，该算法在较少的用户交互情况下，实现了序列图像的种子点位置预判，获得了较满意的图像分割效果。