抄録
We introduce a method for extracting iso-surface from CT data composed of multi-material parts. Most contouring method such as marching cubes assumes that the CT data compose of two materials. However these contouring methods can not be directly applicable for a CT data composed of three materials or more, because of many problems. First is that we need to use n(n-1)/2 iso-value in order to segment a CT data consist of n material. Second is the blurry edges at the junction where more than three materials are adjacent. We propose here a method for segmenting the CT data into its composing materials using a combined method of two well known algorithms in image segmentation, region growing and graph-cut. And we demonstrate the effectiveness of our novel method by constructing high quality polygon models of the segmented parts from CT data.
