Fast and Globally Convergent Pose Estimation from Video Images

By Chien-Ping Lu, Gregory Hager, Eric Mjolsness. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2000

952 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/0f3c956766a250fb3c787cd7d62e45100575da35

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

Revisiting the PnP Problem: A Fast, General and Optimal Solution (2013, methodology): “As a trade off, some direct minimization methods [6,16] minimize instead certain algebraic error functions via local optimization techniques.”
MonoCap: Monocular Human Motion Capture using a CNN Coupled with a Geometric Prior (2017, positive): “e bias of camera rotation estimates due to the adopted weak perspective camera model. To reduce the bias, the camera pose was further reﬁned for each frame using a perspective-n-point (PnP) algorithm [73], given the camera intrinsic parameters provided by the dataset and the 3D/2D structures estimat”
Sparseness Meets Deepness: 3D Human Pose Estimation from Monocular Video (2015, methodology): “To compensate the misalignment, the rotation estimate was refined for each frame with a perspective camera model (the 2D and 3D human pose estimates were fixed) by a perspective-n-point (PnP) algorithm [25]”
Rethinking Reprojection: Closing the Loop for Pose-Aware Shape Reconstruction from a Single Image (2017, methodology): “Third, the pose of the CAD model is determined by annotating coarse landmarks within the natural image that correspond to points on the CAD model - pose is then recovered through a PnP process [15].”
3D Human Pose Estimation from a Single Image via Distance Matrix Regression (2016, methodology): “Additionally, once the 3D pose is estimated, we retrieve the rigid rotation and translation that aligns it with the input image using a PnP algorithm [25].”
Robust Pose Estimation from a Planar Target (2006, background): “From [3], we know that the optimal translation is given by t̂opt ¼ 1 n I 1 n X j V̂j !”
A Review on Video-Based Human Activity Recognition (2013, methodology): “The skeleton silhouette is decomposed into the head, torso and limbs, and the 3D head pose is estimated by finding the rigid transformation, minimizing the sum of square errors with an Orthogonal Iteration (OI) algorithm [122].”
A Robust O(n) Solution to the Perspective-n-Point Problem (2012, background): “Insightful analysis on this subject can be found in Triggs [16].”
3D Human Pose Estimation = 2D Pose Estimation + Matching (2016, methodology): “Such weak-perspective approximations are commonly used to initialize algorithms for perspective (PnP) camera calibration [18], and will be reasonable when the depth variation of the human skeleton is small relative to the overall distance to the camera.”
Vision-Based Pose Estimation From Points With Unknown Correspondences (2014, methodology): “These methods are robust to initial value [2], [4]–[6].”
iLamps: geometrically aware and self-configuring projectors (2003, methodology): “After estimating the external parameters using method in Lu et al. [2000], the error is about 4 pixels.”
Pose Estimation for Augmented Reality: A Hands-On Survey (2016, positive): “…than the best current techniques. markerless model-based tracking It is only slightly less accurate than one the most recent iterative ones (Lu et al. 2000) but much faster and more stable. which do Furthermore, when the output of our al-gorithm is used to initialize a Gauss-Newton…”
Accurate Non-Iterative O(n) Solution to the PnP Problem (2007, methodology): “A natural alternative to non-iterative approaches are iterative ones [19, 5, 14, 17, 20] that rely on minimizing an appropriate criterion.”
StarMap for Category-Agnostic Keypoint and Viewpoint Estimation (2018, background): “It has been well explored by traditional techniques that solve for transformations between corresponding points in the world and image views; this is known as the Perspective-n-Point Problem [13,18].”
Real-time markerless tracking for augmented reality: the virtual visual servoing framework (2006, methodology): “Full-scale, nonlinear optimization techniques (e.g., [35], [36], [12]) consist of minimizing the error between the observation and the forward-projection of the model (this process will be described in detail in Section 2.1).”