Inverse Depth Parametrization for Monocular SLAM

By Javier Civera, Andrew J. Davison, José M. M. Montiel. IEEE Transactions on robotics, 2008

854 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition, Electrical and Electronic Engineering

View Paper: https://www.semanticscholar.org/paper/94743f91cdeab6ced04ef2e405b048feb5456917

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

In Electrical and Electronic Engineering

Citations in the corpus

1-Point RANSAC for extended Kalman filtering: Application to real-time structure from motion and visual odometry (2010, background): “Although the linearization process poses a significant threat to the consistency of the position estimation, it can be elegantly avoided using the inverse depth representation (Montiel et al., 2006; Civera et al., 2008).”
Real-time monocular SLAM: Why filter? (2010, background): “[4]), filtering can immediately give estimates along with uncertainty measures.”
REMODE: Probabilistic, monocular dense reconstruction in real time (2014, background): “In order to deal with higher depth ranges, we base our implementation on the inverse depth [23] and use the currently estimated variance to limit the search for correspondence on the epipolar line.”
Direct Sparse Odometry (2016, background): “In addition to a reduced number of parameters, this naturally enables an inverse depth parametrization, which—in a Gaussian framework— is better suited to represent uncertainty from stereo-based depth estimation, in particular for far-away points [3].”
ORB-SLAM: A Versatile and Accurate Monocular SLAM System (2015, methodology): “These points, which are not seen with sufficient parallax and our system does not include in the map, are very informative of the rotation of the camera [21].”
Estimation, Control, and Planning for Aggressive Flight With a Small Quadrotor With a Single Camera and IMU (2017, methodology): “A. Visual Inertial Odometry The VIO system localizes the rigid body with respect to the inertial frame using accelerometers, gyroscopes and camera sensors.”
Particle Filter SLAM with High Dimensional Vehicle Model (2009, methodology): “Another alternative is to use an inverse depth parametrization for some time as in [ 8 ].”
SVO: Semidirect Visual Odometry for Monocular and Multicamera Systems (2017, methodology): “For each sample, we apply the camera projection model π ( ) to obtain the corresponding pixel coordinate on the curved epipolar line.”
ORB-SLAM3: An Accurate Open-Source Library for Visual, Visual–Inertial, and Multimap SLAM (2020, methodology): “Monocular SLAM was ﬁrst solved in MonoSLAM [13], [14], [53] using an Extended Kalman Filter (EKF) and Shi-Tomasi points that were tracked in subsequent images doing a guided search by correlation.”
High-precision, consistent EKF-based visual-inertial odometry (2013, methodology): “(Civera et al., 2008) showed that the Cartesian-coordinate (XYZ) parametrization results in severely non-Gaussian probability density functions (pdfs) for the features, and degrades accuracy and consistency.”
Stereo DSO: Large-Scale Direct Sparse Visual Odometry with Stereo Cameras (2017, methodology): “The first real-time capable visual SLAM and odometry systems were pioneered around 2000 [2, 24, 5, 3].”
A robust RGB-D SLAM algorithm (2012, methodology): “[10] proposed a feature based approach to improve efficiency and data association in which inverse depth is used in an Extended Kalman Filter (EKF) for feature parameterization.”
ORB-SLAM2: An Open-Source SLAM System for Monocular, Stereo, and RGB-D Cameras (2016, methodology): “, points whose depth cannot be reliably estimated due to little disparity in the stereo camera), using an inverse depth parameterization [6] for the latter.”
OpenVINS: A Research Platform for Visual-Inertial Estimation (2020, background): “The map of environmental landmarks x M contains global 3D positions only for simplicity, while in practice we offer support for different representations (e.g. inverse MSCKF [25], full inverse depth [29], and anchored 3D position [30]).”
Visual SLAM: Why filter? (2012, neutral): “It proved to be useful, especially but not exclusively for monocular SLAM, to represent 3d points using anchored inverse depth coordinates [5].”