Multiple View Geometry in Computer Vision

By Bernhard P. Wrobel. 2001

18,671 citations recorded by the source corpus.

How researchers see this paper

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

In 3D Vision & Pose

Citations in the corpus

Multiview Compressive Coding for 3D Reconstruction (2023, background): “This is framed by the task of 3D reconstruction: given an image, fully reconstruct the scene in 3D. 3D reconstruction is a longstanding problem in AI with applications in robotics and AR/VR. Structure from Motion [18, 60] lifts images to 3D by triangulation.”
Scene Flow to Action Map: A New Representation for RGB-D Based Action Recognition with Convolutional Neural Networks (2017, methodology): “Following the method in [7], we chose a set of matching points in an RGB frame and its corresponding depth map.”
RayNet: Learning Volumetric 3D Reconstruction with Ray Potentials (2018, background): “As a thorough survey on 3D reconstruction techniques is beyond the scope of this paper, we discuss only the most related approaches and refer to [7, 13, 29] for a more thorough review.”
Pixel2Mesh: Generating 3D Mesh Models from Single RGB Images (2018, background): “The former restriction means MVG cannot re-construct unseen parts of the object, and thus it usually takes a long time to get enough views for a good reconstruction; the latter restriction means MVG cannot reconstruct non-lambertian (e.g. reﬂective or transparent) or textureless objects.”
3D Shape Induction from 2D Views of Multiple Objects (2016, methodology): “Structure from motion (SfM) [12] can be used to recover the shape of rigid objects from their views taken from unknown viewpoints by jointly reasoning about point correspondences and camera projections.”
Lifting Object Detection Datasets into 3D (2015, background): “Multiview reconstruction has since been largely solved in the rigid case [8], as calibration parameters and correspondences can be reliably estimated and the problem reduces to a well-understood geometric optimization problem.”
Depth Map Prediction from a Single Image using a Multi-Scale Deep Network (2014, background): “Provided accurate image correspondences, depth can be recovered deterministically in the stereo case [5].”
A Large Dataset to Train Convolutional Networks for Disparity, Optical Flow, and Scene Flow Estimation (2015, methodology): “Since the structure of the prob-Color lem is well known (correspondences can only be found in accordance with the epipolar geometry [9]), we introduced an alternative architecture – the DispNetCorr – in which we explicitly correlate features along horizontal scanlines.”
Unsupervised Learning of Disentangled Representations from Video (2017, methodology): “Classic structure-from-motion methods use an explicit afﬁne projection model to extract a 3D point cloud and camera homography matrices [7].”
The “Something Something” Video Database for Learning and Evaluating Visual Common Sense (2017, background): “These include 3-D geometry (which can reveal itself through multiple views [7]), material properties (such as deformability, elasticity, stiffness, etc.”
Dex-Net 2.0: Deep Learning to Plan Robust Grasps with Synthetic Point Clouds and Analytic Grasp Metrics (2017, methodology): “intrinsics [20] and let Tc be the 3D pose of the camera.”
Gibson Env: Real-World Perception for Embodied Agents (2018, methodology): “depth, is available in the input [39, 58, 18, 78], the pipeline can make use of geometric approaches to be more robust to large viewpoint changes and implausible deformations.”
Generic 3D Representation via Pose Estimation and Matching (2016, methodology): “This is similar to the concept of vanishing points, and from the theoretical perspective, would be intriguing and explain the pose estimation results, since three common vanishing points are theoretically enough for a full angular pose estimation [13,26].”
Kornia: an Open Source Differentiable Computer Vision Library for PyTorch (2019, background): “Multi-view reconstruction is a well understood problem with a good geometric model [21], and many approaches for matching and optimization [39, 8, 51], and some recent promising deep learning approaches [34].”
Improving Image-Based Localization by Active Correspondence Search (2012, methodology): “After finding N =100 matches, the camera pose is estimated with a RANSAC-variant [22] using the 6-point DLT algorithm [16].”
VINS-Mono: A Robust and Versatile Monocular Visual-Inertial State Estimator (2017, methodology): “1) 2-D–2-D: A fundamental matrix test with RANSAC [33].”
Audio-Visual Embodied Navigation (2019, background): “Navigation has a long history in robotics, where a premium is placed on rigorous geometric maps [91, 44].”