THESIS
2011
xi, 37 p. : col. ill. ; 30 cm
Abstract
Mesh segmentation plays an important role in geometry processing and computer graphics applications. It serves as a pre-processing step for many applications such as shape retrieval, skeleton extraction, texture mapping and shape matching. Fully automatic segmentation is still a hard problem as the results may not be consistent with user intention. Thus interactive segmentation remains an interesting topic. Recent work in interactive segmentation focuses on designing good tools that compute a user-desired cut with minimal amount of interactions. These methods require either multiple in-segment strokes or on/cross-boundary stroke with cutting direction specified....[
Read more ]
Mesh segmentation plays an important role in geometry processing and computer graphics applications. It serves as a pre-processing step for many applications such as shape retrieval, skeleton extraction, texture mapping and shape matching. Fully automatic segmentation is still a hard problem as the results may not be consistent with user intention. Thus interactive segmentation remains an interesting topic. Recent work in interactive segmentation focuses on designing good tools that compute a user-desired cut with minimal amount of interactions. These methods require either multiple in-segment strokes or on/cross-boundary stroke with cutting direction specified.
In this thesis, we propose a novel and intuitive click tool for mesh segmentation. Our tool computes a cut using only click locations without a cut direction. The cut direction is automatically determined through a well-designed isoline selection scheme. The candidate cutting boundaries are isolines of harmonic fields. Our scheme selects the best isoline as the cutting boundary. The segmentation results are always smooth and insensitive to noise and tessellation and require no post-processing. To allow more user control, we design an optional cut refinement method that is also driven by clicks. The cut can be refined to pass through more locations specified by clicks. Experimental result shows that our tool is easy-to-use and produces competitive segmentation quality comparing to the state-of-art segmentation tools.
Post a Comment