Elsevier

Computer-Aided Design

Volume 46, January 2014, Pages 120-128
Computer-Aided Design

Geometry seam carving

https://doi.org/10.1016/j.cad.2013.08.024Get rights and content

Highlights

  • Mesh deformation with preservation of salient features.

  • Transfer of the concept of seam carving from images to 3D meshes.

  • Combination of elastic mesh editing with a novel discrete mesh deformation approach.

  • Adaptation of the surface tessellation to the degree of the deformation distortion.

Abstract

We present a novel approach to feature-aware mesh deformation. Previous mesh editing methods are based on an elastic deformation model and thus tend to uniformly distribute the distortion in a least-squares sense over the entire deformation region. Recent results from image resizing, however, show that discrete local modifications such as deleting or adding connected seams of image pixels in regions with low saliency lead to far superior preservation of local features compared to uniform scaling — the image retargeting analog to least-squares mesh deformation. Hence, we propose a discrete mesh editing scheme that combines elastic as well as plastic deformation (in regions with little geometric detail) by transferring the concept of seam carving from image retargeting to the mesh deformation scenario. A geometry seam consists of a connected strip of triangles within the mesh’s deformation region. By collapsing or splitting the interior edges of this strip, we perform a deletion or insertion operation that is equivalent to image seam carving and can be interpreted as a local plastic deformation. We use a feature measure to rate the geometric saliency of each triangle in the mesh and a well-adjusted distortion measure to determine where the current mesh distortion asks for plastic deformations, i.e., for deletion or insertion of geometry seams. Precomputing a fixed set of low-saliency seams in the deformation region allows us to perform fast seam deletion and insertion operations in a predetermined order such that the local mesh modifications are properly restored when a mesh editing operation is (partially) undone. Geometry seam carving hence enables the deformation of a given mesh in a way that causes stronger distortion in homogeneous mesh regions while salient features are preserved much better.

Introduction

The deformation of three-dimensional (3D) models has a wide range of applications in artistic as well as industrial design. Nowadays, the predominant representations of surfaces are triangle meshes which come at high resolution, often acquired using 3D laser scanning, and exhibiting geometric details at various scales. For a deformation technique to be considered useful for editing of such meshes, it is hence crucial that it meets certain requirements: apart from providing visual feedback for interactive application, it should provide easy to control modeling metaphors. Most importantly, it should generate intuitive and predictable deformation results that are physically plausible and aesthetically pleasing. In order to meet these quality requirements, a deformation method has to preserve local characteristics of a surface, i.e., geometric detail or features, under deformation.

This paper presents a novel mesh deformation technique that puts special emphasis on the aspect of feature preservation. Previous mesh editing approaches are mostly based on an elastic deformation model and usually distribute the distortion over the entire deformation region in a least-squares sense. Recent research on image resizing, however, demonstrated that discrete modifications produce results that are far superior to those obtained by applying uniform scaling, which can be considered as the analog in the image processing world to a least-squares deformation in the mesh editing world.

In their work on image seam carving, Avidan and Shamir  [1] insert or delete a connected seam of image pixels in regions with low energy, yielding realistically looking and visually pleasing resizing results. In our work, we transfer the concept of discrete modifications from the image retargeting to the mesh deformation scenario. Our definition of a seam is closely related to the image setting: a geometry seam is a closed and connected path of low-energy triangles and runs through the modeling region of a mesh. Depending on the characteristics of the deformation, geometry seams can be removed from the mesh by collapsing their interior edges and they can be inserted by splitting these edges, which resembles the delete or insert operations in image seam carving. By precomputing a set of low-saliency seams, we can perform deletion and insertion operations at interactive rates. Furthermore, applying the operations in a predetermined order allows us to properly undo previous editing operations and hence to restore the original model. We use a well-established elastic mesh deformation method which we significantly adapt such that it jointly works with this novel plastic and discrete modification scheme which intervenes when the surface distortion exceeds certain thresholds and calls for additional remeshing. This enables editing of a 3D model, thereby distributing the distortion non-homogeneously over the model, and hence causing stronger deformations in low-saliency mesh regions, while features are preserved much better compared to purely elastic mesh editing methods.

Section snippets

Related work

There exists a wide variety of surface deformation techniques in the literature which, in order to position our work, we roughly classify into two categories. First, there are general deformation techniques that allow for general editing operations and that distribute the deformation error over the entire object. Second, there exist structure-aware techniques that perform a structure analysis in a preprocessing phase and then restrict the allowed modifications to application-dependent editing

Geometry seam carving

Our novel mesh editing scheme emulates a physical surface deformation process that supports elastic deformation, i.e., compression and stretching of the material, as well as plastic deformation, which we model as addition and removal of material from the model. We model the elastic deformation using a well-known Laplacian surface editing technique which we significantly augment in order to enable it to jointly work with our new plastic deformation scheme. This scheme adopts the mesh

Results

To illustrate the quality of the surface meshes under deformation, we compare our results with the results obtained by the following base-line reference solution: as our technique combines elastic with a plastic deformation, the reference solution should provide both properties as well. Hence, we obtain our reference solution by using weighted Laplacian editing, but we recompute the Laplace matrix after every frame. We therefore reset the edge weights as λω+(1λ)ω, where ω are the weights

Conclusion

We presented a mesh deformation technique that combines elastic Laplacian-based mesh deformation with plastic mesh modifications. A precomputed set of geometry seams is split or collapsed in those areas of the mesh that undergo strong distortion. Since, per construction, the seams run through low-saliency regions of the mesh, these regions absorb most of the deformation energy, which hence lightens the distortion in regions that exhibit many features. The topological modification resulting from

Acknowledgments

This project was funded by the DFG Cluster of Excellence on Ultra High-Speed Mobile Information and Communication (UMIC), German Research Foundation grant DFG EXC 89. The Octopus model is courtesy of Mark Pauly. The model of the Seahorse and the Knot with Stars were obtained from the AIM@SHAPE repository. The Welsh Dragon mesh model was released by Bangor University, UK, for Eurographics 2011.

References (21)

  • Dominique Bechmann

    Space deformation models survey

    Comput Graph

    (1994)
  • Hiroshi Masuda et al.

    Preserving form features in interactive mesh deformation

    Comput Aided Des

    (2007)
  • Shai Avidan et al.

    Seam carving for content-aware image resizing

    ACM Trans Graph

    (2007)
  • Botsch Mario, Pauly Mark, Gross Markus, Kobbelt Leif. Primo: coupled prisms for intuitive surface modeling. In: Proc....
  • Andrew Nealen et al.

    Physically based deformable models in computer graphics

    Comput Graph Forum

    (2006)
  • Mario Botsch et al.

    On linear variational surface deformation methods

    IEEE Trans Vis Comput Graphics

    (2008)
  • Mark Pauly et al.

    Discovering structural regularity in 3D geometry

    ACM Trans Graph

    (2008)
  • Martin Bokeloh et al.

    Pattern-aware shape deformation using sliding dockers

    ACM Trans Graph

    (2011)
  • Singh Karan, Fiume Eugene. Wires: a geometric deformation technique. In: Proc. of SIGGRAPH’98. 1998. p....
  • Ran Gal et al.

    iWIRES: an analyze-and-edit approach to shape manipulation

    ACM Trans Graph

    (2009)
There are more references available in the full text version of this article.

Cited by (15)

  • Temporally coherent sculpture of composite objects

    2016, Computers and Graphics (Pergamon)
    Citation Excerpt :

    Finally, Laplacian editing [33] can be used to extend the deformation field between a fixed part of the shape being deformed and a handle. An extension of Laplacian editing can be used to preserve details when stretching adaptive meshes [34]. Although such systems can convey a wider range of operations, the performance cost related to updating and solving sparse linear systems usually restricts their use, in the context of solid sculpting, to mesh posing and detail-preserving deformations.

  • Real-time continuous self-replicating details for shape deformation

    2015, Computers and Graphics (Pergamon)
    Citation Excerpt :

    Kraevoy et al. [5] introduce a non-uniform scaling method able to maintain the important features of a 3D model such as circular parts. Dekkers and Kobbelt [6] avoid undesired detail distortion by extending a successful image resizing technique, called seam-carving [7], to surface meshes. Their “geometry seam-carving” approach preserves the shape and size of salient features and redistributes the distortion over the remaining surface regions between the details.

  • Selective Guidance Normal Filter for Geometric Texture Removal

    2021, IEEE Transactions on Visualization and Computer Graphics
  • Seam Carving Based Image Retargeting: A Survey

    2019, 1st International Informatics and Software Engineering Conference: Innovative Technologies for Digital Transformation, IISEC 2019 - Proceedings
View all citing articles on Scopus
View full text