3D Shape Correspondence

Partial 3D Correspondence from Shape Extremities

Yusuf Sahillioğlu and Yücel Yemez
SIGGRAPH Asia (submitted), 2012

 [pdf]  [code and executable]
Input Isometric (or nearly isometric or partially isometric) mesh pair; partially: general partial matching
Output Partial and/or complete correspondences b/w two meshes
Idea Triplets of qualified matches produce votes based on dense matching around

Scale Normalization for Isometric Shape Matching

Yusuf Sahillioğlu and Yücel Yemez
Pacific Graphics (submitted), 2012

 [pdf]  [code and executable]
Input Isometric (or nearly isometric or partially isometric) mesh pair; partially: part matching
Output Partial and/or complete correspondences b/w two meshes
Idea A distortion measure for partial matching based on raw geodesics since their ratios are preserved under isometric deformations

Coarse-to-Fine Isometric Shape Correspondence by Tracking Symmetric Flips

Yusuf Sahillioğlu and Yücel Yemez
Computer Graphics Forum (in revision cycle), 2012

 [pdf]  [code and executable]
Input Isometric (or nearly isometric) mesh pair
Output Correspondence b/w two meshes at desired resolution
Idea Handle the symmetric flip problem by tracking multiple maps

Minimum-Distortion Isometric Shape Correspondence Using EM Algorithm

Yusuf Sahillioğlu and Yücel Yemez
PAMI 2012 (to appear)

 [pdf]  [code and executable]
Input Isometric (or nearly isometric) mesh pair
Output Coarse correspondence b/w evenly-spaced high-curvature sample vertices on two meshes
Idea Alternating minimization of the isometric distortion

Coarse-to-Fine Combinatorial Matching For Dense Isometric Shape Correspondence

Yusuf Sahillioğlu and Yücel Yemez
Computer Graphics Forum (SGP'11), Vol. 30, No. 5, pp. 1461-1470, 2011

 [pdf]  [ppt]  [code and executable]
Input Isometric (or nearly isometric) mesh pair
Output Correspondence b/w two meshes at desired resolution
Idea Recursively subdivide matched patches into smaller patches to be matched. Also, stay in the original 3D Euclidean space at all times

3D Shape Correspondence by Isometry-Driven Greedy Optimization

Yusuf Sahillioğlu and Yücel Yemez
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 453-458, 2010

 [pdf]  [code and executable] [video]
Input Isometric (or nearly isometric) mesh pair
Output Coarse correspondence b/w evenly-spaced sample vertices on two meshes
Idea Initial spectral correspondence is exposed to a greedy refinement

3D Isometric Shape Correspondence Algorithms (in the making)

Yusuf Sahillioğlu
PhD Thesis, Computer Science Dept., Koç University, Turkey, May 2012 (expected)

 [pdf]  [ppt]
Input Isometric (or nearly isometric or partially isometric) mesh pair or mesh sequence.
Output Coarse correspondence, dense correspondence, partial correspondence, or multiple correspondence
Idea towards PhD

3D Reconstruction

Coarse-to-Fine Surface Reconstruction from Silhouettes and Range Data Using Mesh Deformation

Yusuf Sahillioğlu and Yücel Yemez
Computer Vision and Image Understanding (CVIU), Vol. 114, pp. 334-348, 2010

 [pdf] 
Input Real-world object along w/ its multi-view silhouettes and range data
Output Visual hull and cavity-sensitive refined mesh of the object
Idea A scan line of range data brings the visual hull triangle it intersects to the range surface

Triangulation-free 3D Reconstruction from LiDAR Data

Yusuf Sahillioğlu
International Conference on Computer Graphics & Virtual Reality (CGVR), pp. 27-32, 2010

 [pdf]
Input Unorganized, noisy, and dense 3D points acquired by a LiDAR system
Output Low-resolution 2-manifold triangular mesh approximating the LiDAR surface
Idea Guide the deformation w.r.t. the best-fit tangent planes spread over LiDAR points

Shape from Silhouette Using Topology-Adaptive Mesh Deformation

Yücel Yemez and Yusuf Sahillioğlu
Pattern Recognition Letters, Vol. 30, pp. 1198-1207, 2009

 [pdf] 
Input Real-world object along w/ its multi-view silhouettes and range data
Output Visual hull of the object
Idea Use 2D projections on silhouettes for each 3D vertex to guide the topology-adaptive deformation

A Surface Deformation Framework for 3D Shape Recovery

Yusuf Sahillioğlu
MS Thesis, Computer Science Dept., Koç University, Turkey, 2006

 [pdf]   [ppt]
Input Multi-view silhouettes and/or range data of the real-world object to be reconstructed
Output Reconstruction, i.e., visual hull and/or its refined version, as a 2-manifold triangular mesh
Idea Deform bounding sphere in the guidance of silhouettes, and then refine it further w/ scan lines of the range data

Others

3B İzometrik Şekil Eşleme, Yusuf Sahillioğlu and Yücel Yemez, IEEE Sinyal İşleme ve Uygulamaları Kurultayı(SİU), 2010 [pdf] (Best student paper)
 3D Correspondence by Breadth-First Search Frontiers, Yusuf Sahillioğlu, International Conference on Computer Graphics & Virtual Reality (CGVR), pp. 203-207, 2009 [pdf]
 Marching Rhombic Dodecahedra, Yusuf Sahillioğlu and Alireza Entezari, Technical report (Univ. of Florida), 2008 [pdf]  [ppt]  [vids]
 A Surface Deformation Framework for 3D Shape Recovery, Yusuf Sahillioğlu and Yücel Yemez, Lecture notes in Computer Science (MCRS), Vol. 4105, pp. 570-577, 2006 [pdf]
 Çok Kameralı Video Görüntülerinden Yüzey Deformasyonu ile 3B Şekil Geri Çatma ve İzleme, Yusuf Sahillioğlu and Yücel Yemez, SIU, 2006 [pdf]
 Hair Motion Simulation, Yusuf Sahillioğlu and Bülent Özgüç, Int. Symp. on Computer and Inf. Sciences (ISCIS), pp. 126-135, 2004 [pdf]

 Resume [pdf]

Thanks to my friend Emre Ünal for his help on the design of this site.

as i turn to sand, you took me by the hand [SOAD]