Snake-based boundary search for segmentation of 3D polygonal model

Abstract

Many existing 3D model segmentation methods require user input to accurately segment the virtual models. This paper applies snake-based (or contour) approach to automatically search for the boundary between the two connected functional features. The search is guided by the skeleton model. The proposed approach courageously modifies the geometric meshes with respect to each sectional skeleton to ensure the snake meets end to end to form a complete ring. The result of the ring is then interpolated to fit the local geometric feature model. This step eradicates the non-related feature and invalid rings. In the end, the created ring is reverted to its original coordinates so that the meshes are non-distorted. The entire process is done automatically without any user input. The proposed method is compared to two well-known methods: Shape Diameter method and Core Extraction method. The results generated by the proposed method turn out to be more accurate than the two existing methods.