Lucas-Kanade 20 Years On: A Unifying Framework

By Simon Baker, Iain Matthews. International Journal of Computer Vision, 2004

3,506 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/a35561e15d3270d63574905fca6b44b92ab5ace0

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

Citations in the corpus

Automatic Analysis of Facial Affect: A Survey of Registration, Representation, and Recognition (2015, methodology): “Alternatively to landmark-based approaches, generic image registration techniques such as Robust FFT [150] or Lucas-Kanade approaches [8] can also be used.”
A comprehensive survey on point cloud registration (2021, methodology): “Regarding the transformation matrix, there are four kinds of methods: SVD-based [6], Lucas-Kanade (LK) algorithm [4] and Procrustes analysis [22].”
Distribution fields for tracking (2012, background): “This template can consist of the intensity values, gradient information, or other features [3].”
Supervised Descent Method and Its Applications to Face Alignment (2013, methodology): “Parameterized Appearance Models (PAMs), such as Active Appearance Models [11, 14, 2], Morphable Mod- els [6, 19], Eigentracking [5], and template tracking [22, 30] build an object appearance and shape representation by computing Principal Component Analysis (PCA) on a set of manually labeled data.”
Detailed Real-Time Urban 3D Reconstruction from Video (2007, background): “Hence the proposed estimation adds one additional equation (6) to solve to the original KLT tracking equations.”
Vision-Aided Inertial Navigation for Spacecraft Entry, Descent, and Landing (2009, methodology): “The most widely used technique is the Kanade-LucasTomasi (KLT) tracker [38], and several variations of it [39] , [40].”
GMS: Grid-Based Motion Statistics for Fast, Ultra-Robust Feature Correspondence (2017, background): “More generally, our work is related to optical flow [13, 23, 4, 43, 33, 21, 1, 46], point based coherence techniques [48, 18, 29], patch-match based matchers [12] which directly use smoothness to help match estimation.”
Automatic feature localisation with constrained local models (2008, methodology): “[14] have developed a more efficient update scheme which utilises the inverse compositional update algorithm [15].”
Digital image correlation for surface deformation measurement: historical developments, recent advances and future goals (2018, methodology): “Although the equivalence of the classic NR algorithm with forward additive mapping strategy and the recently introduced IC-GN algorithm has been proved in existing studies [184, 185], practical implementation of the two subpixel registration algorithms does involve differences, and therefore leads t”
Ncorr: Open-Source 2D Digital Image Correlation Matlab Software (2015, methodology): “Ncorr uses the Inverse Compositional Gauss-Newton (ICGN) method as the nonlinear optimizer [39, 40].”
Efficient and generalizable statistical models of shape and appearance for analysis of cardiac MRI (2008, methodology): “A novel algorithm for the efficient fitting of 2D AAMs was recently introduced in (Matthews and Baker, 2004).”
SVO: Semidirect Visual Odometry for Monocular and Multicamera Systems (2017, methodology): “Therefore, we model the depth ﬁlter according to [55] with a two dimensional distribution: the ﬁrst dimension is the inverse depth ρ [56], while the second dimension γ is the inlier probability (see bottom rows in Fig.”
A Database and Evaluation Methodology for Optical Flow (2007, methodology): “The simplest gradient descent algorithm is steepest descent [5], which takes steps in the direction of the negative gradient −Global ∂f .”
Survey on RGB, 3D, Thermal, and Multimodal Approaches for Facial Expression Recognition: History, Trends, and Affect-related Applications (2016, background): “This limitation is circumvented in [79], where a 2D model is fit while a 3D one restricts its shape variations.”