zbMATH — the first resource for mathematics

Seed-growing segmentation of 3-D surfaces from CT-contour data. (English) Zbl 1041.68529
Summary: The article presents a layer-based seed-growing technique for the segmentation of 3-D surfaces from computed Tomography (CT) contour data. It falls in an area called reverse engineering for creating a computer aided design model of an industrial or medical object using CT-scanners for dimensional measurement. After pre-processing, a layer-based 3-D contour model is obtained from the 2-D raw CT-images. A seed-growing technique is developed for extracting user-interested surface features from the CT-contour data. After entering a seed contour on an interested layer, contour points belonging to the same surface or feature as the seed contour are segmented from others by growing the seed contour layer by layer, the so-called seed-growing segmentation. The seed contour can be either a closed contour or an open contour depending on the characteristics of the surface feature to be extracted. For a complex object, the segmentation can be performed in two steps. A composite surface is first segmented from the raw CT-contour data. Individual surface features are further segmented from the extracted composite surface contours. An elastic spline and an optimization algorithm are used in the growing process. A 2-D feature point recognition algorithm is incorporated in the open contour growing process.
68U05 Computer graphics; computational geometry (digital and algorithmic aspects)
Full Text: DOI