Parallel and Deterministic Algorithms from MRFs: Surface Reconstruction

By Davi Geiger, Federico Girosi. 1991

Subfields: Computer Vision and Pattern Recognition, 3D Vision & Pose

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

In 3D Vision & Pose

Citations in the corpus

DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs (2016, background): “We note that mean field inference had been extensively studied for traditional image segmentation tasks [54], [55], [56], but these older models were typically limited to short-range connections.”
A unified mixture framework for motion segmentation: incorporating spatial coherence and estimating the number of models (1996, unknown): “[ 6 ]). Since in the EM algorithm, this calculation must be carried out at every iteration one can understand the reluctance of some researchers to use this prior in a segmentation algorithm.”
Exploiting Inference for Approximate Parameter Learning in Discriminative Fields: An Empirical Study (2005, methodology): “It is based on approximating the partition function Z with the Saddle Point Approximation (SPA) [11].”
Deterministic annealing for clustering, compression, classification, regression, and related optimization problems (1998, methodology): “This approach was adopted by various researchers in the fields of graph-theoretic optimization and computer vision [10], [26], [33], [37], [98], [99], [108].”
Markov Random Field Modeling in Image Analysis (2001, methodology): “It is interesting to note an observation made by Geiger and Girosi: "sometimes a finite (3 solution may be more desirable or robust" ((Geiger and Girosi 1991), pages 406-407) where (3 = ~.”
A new algorithm for non-rigid point matching (2000, methodology): “Our approach to non-rigid point matching closely follo ws our earlier w orkon joint estimation of pose and correspondence using the softassign [10] and deterministic annealing [8].”
Iterative reconstruction techniques in emission computed tomography (2006, background): “Although, in general, no global convergence properties are known, favourable behaviour has been observed by several authors (Geiger and Girosi 1991, Gindi et al 1993, Mumcuoglu et al 1994, Yu and Fessler 2002).”
Stereo Matching Using Belief Propagation (2002, methodology): “There are several methods to solve the MAP-MRF problem: simulated annealing [14], Mean-Field annealing [12], the Graduated Non-Convexity algorithm (GNC) [5], and Variational approximation [17].”
A Taxonomy and Evaluation of Dense Two-Frame Stereo Correspondence Algorithms (2001, background): “Traditional approaches associated with regularization and Markov Random Fields include continuation [17], simulated annealing [47, 75, 6], highest confidence first [28], and mean-field annealing [45].”
Stereo correspondence by dynamic programming on a tree (2005, methodology): “Traditional 2D optimization approaches to approximate d include simulated annealing [12], continuation methods [3], mean-field annealing [11].”
A new point matching algorithm for non-rigid registration (2003, unknown): “This process of deterministic annealing is a useful heuristic in a variety of optimization problems [15,17,23].”
Learning Low-Level Vision (1999, unknown): “Researchers have proposed a variety of approximations (Geman and Geman, 1984; Geiger and Girosi, 1991; Jordan, 1998).”
Prior Learning and Gibbs Reaction-Diffusion (1997, background): “Some connections between these interpretations are also observed in [ 12 ], [13] based on effective energy in statistics mechanics.”
Simultaneous truth and performance level estimation (STAPLE): an algorithm for the validation of image segmentation (2004, methodology): “The mean field approximation has been used for constrained surface reconstruction, and its use has been motivated by the fact that it is the minimum variance Bayes estimator of the true field [49].”