An efficient solution to the five-point relative pose problem

By David Nistér. 2003 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2003. Proceedings., 2003

976 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Mechanical Engineering

View Paper: https://www.semanticscholar.org/paper/32989800af8e0aacc238f77f59adc44eec90c592

How researchers see this paper

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

In Computer Vision and Pattern Recognition

In Mechanical Engineering

Citations in the corpus

Modeling the World from Internet Photo Collections (2008, methodology): “The camera parameters for this pair are estimated using Nistér's implementation of the five point algorithm (Nistér 2004), 10 then the tracks visible in the two images are triangulated.”
Pose Estimation for Augmented Reality: A Hands-On Survey (2016, methodology): “The camera is then localized and the global model improved 6 .”
SVO: Semidirect Visual Odometry for Monocular and Multicamera Systems (2017, methodology): “The algorithm is bootstrapped to obtain the pose of the ﬁrst two keyframes and the initial map using the 5-point relative pose algorithm from [58].”
Skeletal graphs for efficient structure from motion (2008, methodology): “Next, we compute a reconstruction for each matching image pair using the ﬁve-point relative pose algorithm of Nist´er [18] inside of a RANSAC loop.”
An Overview to Visual Odometry and Visual SLAM: Applications to Mobile Robotics (2015, methodology): “Full descriptions of different ways to solve the motion estimation using the above approach are provided by [75, 86,119].”
A Survey of Urban Reconstruction (2013, background): “When working with known intrinsic camera settings, the relation is also often referred to as the essential matrix E , which can be determined even from the correspondences of ﬁve points [Nis04].”
Learning to Find Good Correspondences (2017, methodology): “…for estimating the relative camera motion between two images is to use sparse keypoints, such as SIFT [18], to establish an initial set of correspondences, and reject out-liers with RANSAC [10], using the 5-point algorithm [19] or the 8-point algorithm [17] to retrieve the essential matrix.”
Global Fusion of Relative Motions for Robust, Accurate and Scalable Structure from Motion (2013, methodology): “For matching points X and X ′ in images I i and I j respectively, the two-view epipolar constraint can be written The ﬁve-point algorithm of Nistér [23] inserted as minimal solver in a RANSAC procedure robustly estimates the essential matrices extracted, together with the direction t ij , since the…”
Visual Odometry : Part II: Matching, Robustness, Optimization, and Applications (2012, methodology): “Several five-point minimal solvers were proposed later in [38] – [40], but an efficient implementation, based on [39], was found only in 2003 by Nister [41] and later revised in [42].”
Kimera: From SLAM to spatial perception with 3D dynamic scene graphs (2021, methodology): “Table 4 reports the absolute trajectory errors of Kimera when using 5-point RANSAC, 2-point RANSAC, and when using 2-point RANSAC and IMU-aware feature tracking (label: DVIO).”
VINS-Mono: A Robust and Versatile Monocular Visual-Inertial State Estimator (2017, methodology): “In most cases, a visual-only system can bootstrap itself by derived initial values from relative motion methods, such as the eight-point [32] or ﬁve-point [33] algorithms, or estimating the homogeneous matrices.”
Image Matching Across Wide Baselines: From Paper to Practice (2020, methodology): “This is typically approached with iterative sampling schemes based on RANSAC [36], relying on closed-form solutions for pose solving such as the 5-[62], 7-[41] or 8-point algorithm [39].”
ORB-SLAM: A Versatile and Accurate Monocular SLAM System (2015, methodology): “…scene planarity [3], [20] and recover the relative camera pose from a homography using the method of Faugeras et. al [21], or compute an essential matrix [22], [23] that models planar and general scenes, using the ﬁve-point algorithm of Nister [24], which requires to deal with multiple solutions.”
BA-Net: Dense Bundle Adjustment Network (2018, result): “We further compare with conventional geometric (Nister, 2004; Agarwal et al.) and photometric (Engel et al., 2014) BA.”