Equivalence and efficiency of image alignment algorithms

By Simon Baker, Iain Matthews. Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. CVPR 2001, 2001

517 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/cc25324a4d96d7e70bca5084c887db115aa5c7b6

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

Multiple kernel tracking with SSD (2004, background): “We defer at this point to [1] for an in-depth discussion of the general form and structure of f , and, in particular, conditions under which it is possible to optimize these computations.”
Lucas-Kanade 20 Years On: A Unifying Framework (2004, unknown): “We would also like to thank the anonymous IJCV reviewers and the CVPR reviewers of Baker and Matthews (2001) for their feedback.”
Fast, Robust and Accurate Digital Image Correlation Calculation Without Redundant Computations (2013, methodology): “Subsequently, the incremental warp W (ξ; Δp) is inverted and composed with the current estimate W (ξ; p) to get the updated warping function of the deformed subset [20].”
Real-time image-based tracking of planes using efficient second-order minimization (2004, methodology): “Alternatively, there are methods that minimize the sum-of-squared-differences (SSD) between the reference template and the current image using parametric models [9]–[12].”
The subpixel resolution of optical-flow-based modal analysis (2017, methodology): “An inverse composition Gauss-Newton algorithm (introduced by [11]) produces a matching accuracy to the Newton-Raphson at a higher computational efficiency [12].”
The BoneXpert Method for Automated Determination of Skeletal Maturity (2009, background): “It should be noted that the newer development ofAAMcalledinversecompositionalAAMs[ 17 ]arealsobased on shape-driven AAMs.”
Active Appearance Models Revisited (2004, methodology): “The proof that this role reversal step results in an equivalent algorithm can be found in [Baker and Matthews, 2001, Baker and Matthews, 2003].”
Evaluation of Interest Point Detectors and Feature Descriptors for Visual Tracking (2011, methodology): “4. The position of each ball individually is refined using “inverse-compositional” image alignment ( Baker and Matthews 2001 ) with 3 degrees of freedom (x, y, scale) between the current and the previous frame, and the homography between the current image and a canonical reference frame is computed ”
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”
Efficient and generalizable statistical models of shape and appearance for analysis of cardiac MRI (2008, methodology): “The inverse compositional algorithm There is a wealth of literature on image alignment algorithms and the reader is referred to Baker et al. (2004), Baker and Matthews (2001), Maintz and Viergever (1998), Zitova and Flusser (2003) and the references therein for an overview.”
The template update problem (2004, unknown): “translations, affine warps, and 2D similarity transformations, [1]”
2D vs. 3D Deformable Face Models: Representational Power, Construction, and Real-Time Fitting (2007, methodology): “In Baker and Matthews (2001) and Matthews and Baker (2004) we proposed a real-time algorithm for fitting 2D Active Appearance Models.”
Fast Optical Flow Using Dense Inverse Search (2016, methodology): “Inspired by the inverse compositional image alignment of Baker and Matthews [3, 1] we devise our inverse search procedure (explained in § 2.1) for fast mining of a grid of patch-based correspondences between two input images.”
3-D active appearance models: segmentation of cardiac MR and ultrasound images (2002, methodology): “[24] S. Baker and I. Matthews, “Equivalence and efficiency of image alignment algorithms,” inProc.”
Homography-based 2D Visual Tracking and Servoing (2007, methodology): “…algorithms that approximate the current Jacobian J 0 (which varies from one iteration to another) by a constant Jacobian (Hager and Belhumeur 1998 Baker and Matthews 2001): J 0 J (40) This makes the algorithm faster since the matrix J 0 and the inverse of the matrix S are computed once for all.”
Parallel Tracking and Mapping for Small AR Workspaces (2007, methodology): “We use the inverse composi tional approach of [1], minimising over translation and mean patch intensity difference.”
Statistical shape models for 3D medical image segmentation: A review (2009, unknown): “, 1998) (which only update pose and shape by residuals, while texture is updated directly), compositional updating (Baker and Matthews, 2001) (using thin-plate splines for the pose update) and Direct AAMs (Hou et al.”