Surface and Curve Skeletonization of Large 3D Models on the GPU

Abstract

Abstract-We present a GPU-based framework for extracting surface and curve skeletons of 3D shapes represented as large polygonal meshes. We use an efficient parallel search strategy to compute point-cloud skeletons and their distance and feature transforms with user-defined precision. We regularize skeletons by a new GPU-based geodesic tracing technique which is orders of magnitude faster and more accurate than comparable techniques. We reconstruct the input surface from skeleton clouds using a fast and accurate image-based method. We also show how to reconstruct the skeletal manifold structure as a polygon mesh and the curve skeleton as a polyline. Compared to recent skeletonization methods, our approach offers two orders of magnitude speed-up, high precision, and low memory footprints. We demonstrate our framework on several complex 3D models.

Probabilistic View-based 3D Curve Skeleton Computation on the GPU

“…It is not clear to us why this step is required (or beneficial), as it actually changes the topology of the skeleton, and thus may impair operations such as shape analysis or matching. Properties: Our method maintains all of the desirable properties of curve skeletons advocated by related work (Cornea et al, 2007;Au et al, 2008;Tagliasacchi et al, 2012;Livesu et al, 2012;Jalba et al, 2012): Our skeletons are thin and locally centered within the object. Higher-genus objects (with tunnels) are handled well (see rabbit and rotor models, Fig.…”

“…Curve skeleton extraction has received increased attention in the last years (Dey and Sun, 2006;Reniers et al, 2008;Jalba et al, 2012;Au et al, 2008;Ma et al, 2012;Tagliasacchi et al, 2009;Cao et al, 2010a;Tagliasacchi et al, 2012). All these methods work in object space, i.e.…”

Improved Part-Based Segmentation of Voxel Shapes by Skeleton Cut Spaces

“…Chang et al compute shape medial surfaces, separate their manifolds, and back project each manifold on the shape surface to nd a segment [8]. A similar segmentation method, using high-resolution point-cloud skeletons computed on the GPU [17] along the method of Reniers et al [40], is proposed in [22]. Similar high-resolution surface and curve skeletons can be computed in voxel space using an advection model [18].…”

“…[3,18,40,41,46], as long as it outputs regularized skeletons. This makes our method directly applicable to mesh-based shapes, which allow fast medial-surface extraction [17], without the additional cost of voxelization.…”

An Unified Multiscale Framework for Planar, Surface, and Curve Skeletonization

“…Other methods use edge collapses [38], starting from a mesh segmentation [32]. Surface skeletons can be extracted from oriented point clouds [29], [41] or polygon meshes [36], [45] by searching for maximally inscribed balls tangent at at least two shape points. Curve skeletons can be extracted from point clouds as centers of cloud projections on a cut plane which optimizes for circularity [68].…”