Feature Correspondence Via Graph Matching: Models and Global Optimization

By Lorenzo Torresani, Vladimir Kolmogorov, Carsten Rother. European Conference on Computer Vision, 2008

440 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/695465a300d7d6387ad5c59054873bf40d5841f4

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

Fast Exact Hyper-graph Matching with Dynamic Programming for Spatio-temporal Data (2014, neutral): “This work thus falls into the category of methods calculating the exact solution for an approximated model, unlike methods calculating an approximate solution, for instance [5,14,30,48,52,58,59] or methods which perform exact structure preserving matching like [29].”
Deformable Medical Image Registration: A Survey (2013, methodology): “also employed pairwise constraints to model local spatial coherence [289].”
Bayesian Inference of Bijective Non-Rigid Shape Correspondence (2016, methodology): “Minimumdistortion methods establish the matching by minimizing some structure distortion, which can include similarity of local features [30, 11, 6, 44], geodesic [27, 10, 12] or diffusion distances [14], or a combination thereof [40].”
Mining And-Or Graphs for Graph Matching and Object Discovery (2015, methodology): “Most of them [3, 2, 18, 30] are supervised methods, i.”
Learning shape correspondence with anisotropic convolutional neural networks (2016, methodology): “Minimum-distortion methods establish the matching by minimizing some structure distortion, which can include similarity of local features [OMMG10,ASC11,ZBVH09], geodesic [MS05,BBK06,CK15] or diffusion distances [CLL ∗ 05], or a combination thereof [TKR08], or higher-order structures [Z ∗ 10].”
Attributed Graph Mining and Matching: An Attempt to Define and Extract Soft Attributed Patterns (2014, methodology): “Views of graph matching: Given a graph template and a number of ARGs, conventional algorithms for learning graph matching [4, 16, 17, 24] mainly train matching parameters, and Cho et al.”
Product Manifold Filter: Non-rigid Shape Correspondence via Kernel Density Estimation in the Product Space (2017, methodology): “Minimum-distortion methods establish the matching by minimizing some structure distortion, which can include similarity of local features [32, 14, 8, 45], geodesic [29, 13, 15] or diffusion distances [17], or a combination thereof [41].”
Structured Learning and Prediction in Computer Vision (2011, background): “Feature matching between images has been addressed using graph matching in [150], where dual decomposition provided strong relaxations.”
Reweighted Random Walks for Graph Matching (2010, methodology): “In this section, we performed feature point matching on the CMU House sequence which has been widely used in previous works [7,5] and compared with other methods.”
Image Co-segmentation via Consistent Functional Maps (2013, background): “Since the graph matching problem is NP-hard in most versions, an important topic under this theme is to design efficient algorithms for approximately solving the assignment problem [7, 23].”
Category Modeling from Just a Single Labeling: Use Depth Information to Guide the Learning of 2D Models (2013, methodology): “Conventional algorithms for learning graph matching [2, 3, 4] have focused on training the weights of different graphical attributes, given a template graph (the category model) and multiple target graphs.”
Factorized Graph Matching (2016, methodology): “[40] designed a complex objective function which can be efficiently optimized by dual decomposition.”
Finding Matches in a Haystack: A Max-Pooling Strategy for Graph Matching in the Presence of Outliers (2014, methodology): “In practice, this is comparable or faster in speed than the recent state-of-the-art algorithms [6, 20, 25, 32].”
SuperGlue: Learning Feature Matching With Graph Neural Networks (2019, background): “For local features, the computer vision literature of the 2000s [4, 24, 51] uses handcrafted costs with many heuristics, making it complex and brittle.”
Improving Data Association by Joint Modeling of Pedestrian Trajectories and Groupings (2010, methodology): “For the inference, we use Dual Decomposition (DD) [23], building on the code made available by [24].”
Randomized Optimum Models for Structured Prediction (2012, background): “Some notable examples include the problem of modeling correspondences in computer vision (Torresani et al., 2008); the assignment of workers to jobs in operations research; or computing word alignment in natural language processing (Taskar et al.”
MRF Energy Minimization and Beyond via Dual Decomposition (2011, positive): “We refer the interested reader to [48] for a nice application of the dual decomposition framework in this regard (capable of obtaining globally optimal solutions to the graph matching problem).”