Continuous Global Optimization in Multiview 3D Reconstruction

Kolev, Kalin; Klodt, Maria; Brox, Thomas; Esedoḡlu, Selim; Cremers, Daniel

doi:10.1007/978-3-540-74198-5_34

Cited by 42 publications

(47 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Alternatively, one can dynamically integrate silhouette-aligning forces in each optimization step of stereo reconstruction. Kolev et al [22] used a regional term on volumetric grids to enforce the silhouette constraint, which is similar to the Chan-Vese model in 2D/3D image segmentation [23,24]. This regional term can be updated during energy optimization, so the occluded regions can be determined based on currently estimated surface.…”

Section: Related Workmentioning

confidence: 99%

Multiview stereo and silhouette fusion via minimizing generalized reprojection error

Wang

Jia

et al. 2015

Image and Vision Computing

View full text Add to dashboard Cite

Accurate reconstruction of 3D geometrical shape from a set of calibrated 2D multiview images is an active yet challenging task in computer vision. The existing multiview stereo methods usually perform poorly in recovering deeply concave and thinly protruding structures, and suffer from several common problems like slow convergence, sensitivity to initial conditions, and high memory requirements. To address these issues, we propose a two-phase optimization method for generalized reprojection error minimization (TwGREM), where a generalized framework of reprojection error is proposed to integrate stereo and silhouette cues into a unified energy function. For the minimization of the function, we first introduce a convex relaxation on 3D volumetric grids which can be efficiently solved using variable splitting and Chambolle projection. Then, the resulting surface is parameterized as a triangle mesh and refined using surface evolution to obtain a high-quality 3D reconstruction. Our comparative experiments with several state-of-the-art methods show that the performance of TwGREM based 3D reconstruction is among the highest with respect to accuracy and efficiency, especially for data with smooth texture and sparsely sampled viewpoints.

show abstract

Section: Related Workmentioning

confidence: 99%

Multiview stereo and silhouette fusion via minimizing generalized reprojection error

Wang

Jia

et al. 2015

Image and Vision Computing

View full text Add to dashboard Cite

show abstract

“…Global methods typically define a global cost function for shapes and then use some global optimization algorithm to recover a shape that minimizes the cost function. Examples of global methods include approaches based on volumetric Markov Random Field (MRF) models [3], which utilize graph cut optimization techniques [2], and variational approaches, which use convex optimization [28]. A common property of global methods is a relatively large memory usage and time complexity due to the use of a dense volumetric grid.…”

Section: General Overviewmentioning

confidence: 99%

Multi-View Surface Reconstruction by Quasi-Dense Wide Baseline Matching

Kannala¹,

Ylimäki²,

Koskenkorva³

et al. 2011

Emerging Topics in Computer Vision and Its Applications

View full text Add to dashboard Cite

This chapter provides a review on correspondence growing techniques which have been used in multi-view stereo reconstruction problems. Typically these methods approach the problem of multi-view image matching by first determining a sparse set of feature correspondences between pairs of views and then iteratively expanding the matching regions. Sometimes such techniques are also referred by terms like match propagation, quasi-dense matching, or surface growing. Besides providing an overview of the research area, this chapter introduces a particular method, called quasi-dense wide baseline matching, which employs the best-first correspondence growing principle for matching pixels in views with substantially different viewpoints. In addition, the properties and performance of different methods are illustrated by examples and experiments with real images.

show abstract

“…Variational approach has also been successfully applied in binocular scene flow scenario, in which depth and motion are jointly optimized. In MVS area, Kolev [14] proposed a continuous global optimization algorithm parallel to discrete volumetric graph cuts [28]. Although it is based on variational technique, its solution is in the 3D space, while ours is based on the 2D image space.…”

Section: Related Workmentioning

confidence: 99%

Continuous depth estimation for multi-view stereo

Liu

Cao

Dai

et al. 2009

2009 IEEE Conference on Computer Vision and Pattern Recognition

View full text Add to dashboard Cite

Depth-map merging approaches have become more and more popular in multi-view stereo (MVS) because of their flexibility and superior performance. The quality of depth map used for merging is vital for accurate 3D reconstruction. While traditional depth map estimation has been performed in a discrete manner, we suggest the use of a continuous counterpart. In this paper, we first integrate silhouette information and epipolar constraint into the variational method for continuous depth map estimation. Then, several depth candidates are generated based on a multiple starting scales (MSS) framework. From these candidates, refined depth maps for each view are synthesized according to path-based NCC (normalized cross correlation) metric. Finally, the multiview depth maps are merged to produce 3D models. Our algorithm excels at detail capture and produces one of the most accurate results among the current algorithms for sparse MVS datasets according to the Middlebury benchmark. Additionally, our approach shows its outstanding robustness and accuracy in free-viewpoint video scenario.

show abstract

Continuous Global Optimization in Multiview 3D Reconstruction

Cited by 42 publications

References 32 publications

Multiview stereo and silhouette fusion via minimizing generalized reprojection error

Multiview stereo and silhouette fusion via minimizing generalized reprojection error

Multi-View Surface Reconstruction by Quasi-Dense Wide Baseline Matching

Continuous depth estimation for multi-view stereo

Contact Info

Product

Resources

About