Prop-cut: A mesh cutting method based on Tikhonov regularization

We present an interactive mesh cutting method in this paper, which is based on a formulation of semi-supervised learning. Users first assign foreground and background seed faces on a given triangular mesh by sketch lines. Then the system computes a scalar field across the mesh. After that a coarse segmentation boundary is computed out respect to a specific iso-value, which leads to a refined boundary by tracing the isoline in the scalar field. Our proposed methods are easy for implementing. The presented computing framework can not only do segmentation for single static mesh models using shape information, but also do segmentation for dynamic mesh models based on deformation information. By integrating the proposed mesh cutout tool, we also demonstrate a simple sketch-based mesh editing system. In our system the cutting results can be further deformed, morphed, or cut-and-pasted.