26,718 Hits in 6.2 sec

Out-of-core Resampling of Gigantic Point Clouds

Arnaud Bletterer, Frédéric Payan, Marc Antonini, Anis Meftah
2018 Eurographics Symposium on Geometry Processing  
In this work, we introduce a local graph-based structure that enables to manipulate gigantic point clouds, by taking advantage of their inherent structure.  ...  However, the amount of points acquired (several billions) and their distribution raise the problem of sampling a surface optimally.  ...  However, these point clouds exhibit numerous defects in terms of sampling quality (overlapping regions, highly non-uniform distributions, etc.), and contain too many samples to be processed as they are  ... 
doi:10.2312/sgp.20181178 dblp:conf/sgp/BlettererPAM18 fatcat:jn5zyldfi5ewfgeyuqlp2dpfga

Towards the Reconstruction of Wide Historical Sites: A Local Graph-based Representation to Resample Gigantic Acquisitions

Arnaud Bletterer, Frédéric Payan, Marc Antonini, Anis Meftah
2018 Eurographics Workshop on Graphics and Cultural Heritage  
In this paper, we propose a local graph-based structure to deal with the set of LiDAR acquisitions of a digitization campaign.  ...  Those local graphs are then connected together to obtain a single and global representation of the original scene. This structure is particularly suitable for resampling gigantic points clouds.  ...  Thus, they exhibit numerous defects in terms of sampling quality (highly non-uniform distributions, noise, etc.), and may contain too many samples to be processed as they are.  ... 
doi:10.2312/gch.20181342 dblp:conf/vast/BlettererPAM18 fatcat:2kkedqdrevhdhjb5oj7jp36squ

Quad-tree partitioned compressed sensing for depth map coding

Ying Liu, Krishna Rao Vijayanagar, Joohee Kim
2014 2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)  
At the decoder, pixel domain totalvariation minimization is performed for high quality depth map reconstruction. Experiments presented herein illustrate and support these developments.  ...  I frame and each of the subsequent residual frames.  ...  On the other hand, graph based transform (GBT) has also been proposed for CS based depth map coding [14] .  ... 
doi:10.1109/icassp.2014.6853721 dblp:conf/icassp/LiuVK14 fatcat:wog6zbmdjjdilcmzyo4zdasdyu

DEMEA: Deep Mesh Autoencoders for Non-Rigidly Deforming Objects [article]

Edgar Tretschk, Ayush Tewari, Michael Zollhöfer, Vladislav Golyanik, Christian Theobalt
2020 arXiv   pre-print
We demonstrate multiple applications of DEMEA, including non-rigid 3D reconstruction from depth and shading cues, non-rigid surface tracking, as well as the transfer of deformations over different meshes  ...  Mesh autoencoders are commonly used for dimensionality reduction, sampling and mesh modeling.  ...  We have shown multiple applications of our architecture including non-rigid reconstruction from real depth maps and 3D reconstruction of textureless surfaces from images.  ... 
arXiv:1905.10290v2 fatcat:r4lrju4zwbau3gvsvbsvqu3iyu

Depth map reconstruction using color-based region merging

Camilo Dorea, Ricardo L. de Queiroz
2011 2011 18th IEEE International Conference on Image Processing  
A region-based framework is introduced wherein a color-based partition of the image is created and depth uncertainty areas are identified according to the alignment of detected depth discontinuities and  ...  A color-based homogeneity criterion is used to guide a constrained region merging process and reconstruct depth estimates within the uncertainty areas.  ...  Fig. 4 : 4 Fig. 4: Detail crop of (a) color-homogeneous regions of uniform depth outside of uncertainty mask and (b) reconstructed depth map.  ... 
doi:10.1109/icip.2011.6115861 dblp:conf/icip/DoreaQ11 fatcat:to7c7obxprcedfgfru37kh5isa

Edge-adaptive depth map coding with lifting transform on graphs

Yung-Hsuan Chao, Antonio Ortega, Wei Hu, Gene Cheung
2015 2015 Picture Coding Symposium (PCS)  
We present a novel edge adaptive depth map coding based on lifting on graphs.  ...  Experiments show that the optimized sampling approach achieves better results than the conventional maximum cut based splitting in terms of transform efficiency and reconstruction quality.  ...  The input depth maps are first divided into non- Fig. 3 : 3 Depth map encoder transforms most frequently used in the training images are precomputed off-line for simple lookup.  ... 
doi:10.1109/pcs.2015.7170047 dblp:conf/pcs/ChaoOHC15 fatcat:irg5mg7lgbf4ti4ok4otw4vlzq

DeepDeform: Learning Non-rigid RGB-D Reconstruction with Semi-supervised Data [article]

Aljaž Božič, Michael Zollhöfer, Christian Theobalt, Matthias Nießner
2020 arXiv   pre-print
Based on this corpus, we introduce a data-driven non-rigid feature matching approach, which we integrate into an optimization-based reconstruction pipeline.  ...  Applying data-driven approaches to non-rigid 3D reconstruction has been difficult, which we believe can be attributed to the lack of a large-scale training corpus.  ...  Acknowledgements We would like to thank the expert annotators Sathya Ashok, Omar Hedayat, Haya Irfan, Azza Jenane, Soh Yee Lee, Suzana Spasova, and Weile Weng for their efforts in building the dataset.  ... 
arXiv:1912.04302v2 fatcat:7t2d3xekyvey3doiszgnjqkwe4

A Practical Approach for Super-Resolution using Photometric cue and Graph Cuts

S. Sharma, M. V. Joshi
2007 Procedings of the British Machine Vision Conference 2007  
We model the high resolution structure (surface gradients) as a Markov Random Field (MRF) and use graph cuts with discontinuity preservation to get a high resolution depth map.  ...  In this paper, we propose an approach to obtain super-resolved image and super-resolved depth map using photometric cue.  ...  André Jalobeanu, LSIIT, Université de Louis Pasteur, Strasbourg, France for his constructive suggestions and comments.  ... 
doi:10.5244/c.21.83 dblp:conf/bmvc/SharmaJ07 fatcat:rptsec5l6fafjpfaagg7tkny5q

DeepDeform: Learning Non-Rigid RGB-D Reconstruction With Semi-Supervised Data

Aljaz Bozic, Michael Zollhofer, Christian Theobalt, Matthias NieBner
2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
With this data, we propose a new method for non-rigid matching, which we integrate into a non-rigid reconstruction approach.  ...  Figure 1 : We propose a semi-supervised strategy combining self-supervision with sparse annotations to build a large-scale RGB-D dataset of non-rigidly deforming scenes (400 scenes, 390,000 frames, 2,537  ...  Grant 4DRepLy, and the German Research Foundation (DFG) Grant Making Machine Learning on Static and Dynamic 3D Data Practical.  ... 
doi:10.1109/cvpr42600.2020.00703 dblp:conf/cvpr/BozicZTN20 fatcat:xugfdxa7wjelvhumz3p4ssuc7u

Online Reconstruction of Indoor Scenes from RGB-D Streams

Hao Wang, Jun Wang, Liang Wang
2016 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
A system capable of performing robust online volumetric reconstruction of indoor scenes based on input from a handheld RGB-D camera is presented.  ...  Our system is powered by a two-pass reconstruction scheme. The first pass tracks camera poses at video rate and simultaneously constructs a pose graph on-the-fly.  ...  Since the reconstructed map and refined trajectories are from asynchronous threads, a non-rigid map optimization is coupled with incremental pose graph optimization to obtain consistent results.  ... 
doi:10.1109/cvpr.2016.356 dblp:conf/cvpr/WangWW16 fatcat:bpokcmaapff3dkw7cjylkc5gce

Low-Cost Real-Time 3D Reconstruction of Large-Scale Excavation Sites using an RGB-D Camera [article]

Michael Zollhöfer, Christian Siegl, Bert Riffelmacher, Mark Vetter, Boris Dreyer, Marc Stamminger, Frank Bauer
2014 Eurographics Workshop on Graphics and Cultural Heritage  
After a raw reconstruction has been acquired, we interactively warp the digital model to fit a geo-referenced map using a handle based deformation paradigm.  ...  The quality of the acquired digitized raw 3D models is evaluated by comparing them to actual imagery and a geo-referenced map of the excavation site.  ...  The deformation nodes of the proxy deformation graph G are com- puted via Poisson Disc Sampling.  ... 
doi:10.2312/gch.20141298 fatcat:av5mb2v2prbjrp4kgxdf57rm4y

Deformation-based loop closure for large scale dense RGB-D SLAM

Thomas Whelan, Michael Kaess, John J. Leonard, John McDonald
2013 2013 IEEE/RSJ International Conference on Intelligent Robots and Systems  
By combining pose graph optimisation with non-rigid deformation of a dense map we are able to obtain highly accurate dense maps over large scale trajectories that are both locally and globally consistent  ...  Central to this work is the use of an "as-rigid-aspossible" space deformation for efficient dense map correction in a pose graph optimisation framework.  ...  the Irish Research Council and by ONR grants N00014-10-1-0936, N00014-11-1-0688, N00014-12-1-0093, and N00014-12-10020.  ... 
doi:10.1109/iros.2013.6696405 fatcat:smolmv6cjzbaho6red2vw6eqlm

Spatiotemporal Texture Reconstruction for Dynamic Objects Using a Single RGB‐D Camera

Hyomin Kim, Jungeon Kim, Hyeonseo Nam, Jaesik Park, Seungyong Lee
2021 Computer graphics forum (Print)  
The input of our framework is a 3D template model and an RGB-D image sequence.  ...  We formulate the problem as an MRF optimization and define cost functions to reconstruct a plausible spatiotemporal texture for a dynamic object.  ...  To obtain a deformation sequence of the template mesh M that matches the input depth stream, we adopt a non-rigid registration scheme based on an embedded deformation (ED) graph [SSP07] .  ... 
doi:10.1111/cgf.142652 fatcat:iuxmvyimgfgergyopnred2bfd4

gradSLAM: Automagically differentiable SLAM [article]

Krishna Murthy Jatavallabhula, Soroush Saryazdi, Ganesh Iyer, Liam Paull
2020 arXiv   pre-print
This amalgamation of dense SLAM with computational graphs enables us to backprop all the way from 3D maps to 2D pixels, opening up new possibilities in gradient-based learning for SLAM.  ...  Blending representation learning approaches with simultaneous localization and mapping (SLAM) systems is an open question, because of their highly modular and complex nature.  ...  Application: RGB and depth completion In Fig. 13 , we similarly introduce such occluders (top row) and pixel noise (bottom row) in one of the depth maps of a sequence and reconstruct the scene using ∇  ... 
arXiv:1910.10672v3 fatcat:y6aoujprevbs3k72lve6bmttqa

Seeing Beyond Foreground Occlusion: A Joint Framework for SAP-Based Scene Depth and Appearance Reconstruction

Zhaolin Xiao, Lipeng Si, Guoqing Zhou
2017 IEEE Journal on Selected Topics in Signal Processing  
Assuming both scene depth and appearance are unknown, we reconstruct 3D scene from camera array data by selecting optimal views with pixel based clustering.  ...  Even when all views are partially occluded, our approach can recover accurate depth map as well as scene appearance using camera array data.  ...  (c) Recovered depth map using entropy based method [43] . (d) Recovered depth map using our method.  ... 
doi:10.1109/jstsp.2017.2715012 fatcat:3duij3q4kvcptg5jr5tlqhqdyi
« Previous Showing results 1 — 15 out of 26,718 results