Real-Time Visibility-Based Fusion of Depth Maps

By Paul C. Merrell, Amir Akbarzadeh, Liang Wang, Philippos Mordohai, Jan-Michael Frahm, Ruigang Yang, David Nistér, Marc Pollefeys. 2007

420 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Aerospace Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

SurfaceNet: An End-to-End 3D Neural Network for Multiview Stereopsis (2017, methodology): “In available multiview stereo pipelines, sparse features are first detected and then propagated to a dense point cloud for covering the whole surface [8, 11], or multiview depth maps are first computed followed with a depth map fusion step to obtain the 3D reconstruction of the object [15, 17].”
Learning a Multi-View Stereo Machine (2017, methodology): “These depth maps can then be fused using a variety of different techniques [38, 8, 33, 59, 30].”
Multi-View Stereo: A Tutorial (2015, methodology): “There exist many different techniques that merge multiple depthmaps [52, 76, 94, 142, 153, 207].”
Using Multiple Hypotheses to Improve Depth-Maps for Multi-View Stereo (2008, methodology): “Recent work has demonstrated that depth-map estimation and integration paradigm may be used to produce accurate results with greatly reduced computation time [18] or real-time [17].”
Continuous depth estimation for multi-view stereo (2009, background): “As for works that concentrate on multi-view depth maps merging, Merrell [16] addresses the problem of real-time depth map merging via GPU technique .”
Detailed Real-Time Urban 3D Reconstruction from Video (2007, methodology): “Details an extensions of our stereo fusion algorithm are given in [37].”
Live dense reconstruction with a single moving camera (2010, background): “depth map fusion [6] are combined to reconstruct large urban models.”
Ensemble Learning for Confidence Measures in Stereo Vision (2013, background): “Applications where accurate stereo confidence measures are essential in raising reliability of computer vision include sparse [19] or dense [16] 3D scene reconstructions.”
High-quality single-shot capture of facial geometry (2010, methodology): “Outliers are removed using a simplified approach of [Merrell et al. 2007].”
MVSNet: Depth Inference for Unstructured Multi-view Stereo (2018, methodology): “The visibility-based fusion algorithm [26] is used in our reconstruction, where depth occlusions and violations across different viewpoints are minimized.”
Robust Probabilistic Occupancy Grid Estimation from Positive and Negative Distance Fields (2012, methodology): “[24] to obtain a probability mass function for depth given NCC scores, assuming a uniform depth prior.”
BlendedMVS: A Large-Scale Dataset for Generalized Multi-View Stereo Networks (2019, methodology): “For post-processing, we also follow previous works [34, 35] to apply the visibility-based depth map fusion [20], average depth map fusion and visibility depth map ﬁlter to generate the 3D point cloud.”
Consistent Depth Maps Recovery from a Video Sequence (2009, background): “Our experiments demonstrate that it is rather effective to improve the depth estimation around the occlusion boundaries and in the textureless regions.”
High Accuracy and Visibility-Consistent Dense Multiview Stereo (2012, positive): “While visibility is taken into account to fuse depth maps in [45], the focus on high performance prevents the use of a global optimization.”
NeuS: Learning Neural Implicit Surfaces by Volume Rendering for Multi-view Reconstruction (2021, methodology): “Point- and surface-based reconstruction methods estimate the depth map of each pixel by exploiting inter-image photometric consistency [8] and then fuse the depth maps into a global dense point cloud [25, 48].”
Segmentation and Recognition Using Structure from Motion Point Clouds (2008, background): “The structure from motion, or SfM, community [1] has demonstrated the value of ego-motion derived data, and their modeling efforts have even extend to stationary geometry of cities [2].”
Multiview Neural Surface Reconstruction by Disentangling Geometry and Appearance (2020, methodology): “However, a post-processing steps of depth fusion [6, 36] followed by the Poisson Surface Reconstruction algorithm [24] are required for producing a valid 3D watertight surface reconstruction.”