Simple Reconstruction of Tree Branches from a Single Range Image

3D modeling of trees in real environments is a challenge in computergraphics and computer vision, since the geometric shape and topologicalstructure of trees are more complex than conventional artificialobjects. In this paper, we present a multi-process approach that ismainly performed in 2D space to faithfully construct a 3D model of thetrunk and main branches of a real tree from a single range image. Therange image is first segmented into patches by jump edge detection basedon depth discontinuity. Coarse skeleton points and initial radii arethen computed from the contour of each patch. Axis directions areestimated using cylinder fitting in the neighborhood of each coarseskeleton point. With the help of axis directions, skeleton nodes andcorresponding radii are computed. Finally, these skeleton nodes arehierarchically connected, and improper radii are modified based on plantknowledge. 3D models generated from single range images of real treesdemonstrate the effectiveness of our method. The main contributions ofthis paper are simple reconstruction by virtue of image storage orderof single scan and skeleton computation based on axis directions.