Energy Minimization Methods in Computer Vision and Pattern Recognition
8th International Conference, EMMCVPR 2011, St. Petersburg, Russia, July 25-27, 2011, Proceedings
Published on 22. July 2011
450 pages
978-3-642-23094-3 (ISBN)
Description
This book constitutes the refereed proceedings of the 8th International Workshop on Energy Minimization Methods in Computer Vision and Pattern Recognition, EMMCVPR 2011, held in St. Petersburg, Russia in July , 2011. The book presents 30 revised full papers selected from a total of 52 submissions. The book is divided in sections on discrete and continuous optimization, segmentation, motion and video, learning and shape analysis.
More details
Other editions
Additional editions
Yuri Boykov | Fredrik Kahl | Victor Lempitsky
Energy Minimization Methods in Computer Vision and Pattern Recognition
8th International Conference, EMMCVPR 2011, St. Petersburg, Russia, July 25-27, 2011, Proceedings
Book
07/2011
1st Edition
Springer
€53.49
Shipment within 7-9 days
Content
- Title
- Preface
- Table of Contents
- Discrete Optimization
- A Distributed Mincut/Maxflow Algorithm Combining Path Augmentation and Push-Relabel
- Introduction
- Mincut and Push-Relabel
- Region Discharge Revisited
- Augmented Path Region Discharge
- New Distance Function
- New Region Discharge
- Complexity of the Sequential ARD
- Experiments
- Synthetic Instances
- Real Instances
- Conclusion
- References
- Minimizing Count-Based High Order Terms in Markov Random Fields
- Introduction
- Count-Based Terms as Integer Linear Programs
- Integer Programming Formulation
- Special Cases
- Composite Features
- Optimization Strategies
- Linear Programming
- Customized Cutting Planes
- Branch-and-Cut
- Experiments
- Histogram-Based Image Segmentation
- Binary Texture Denoising
- Completion of Binary Textures
- Conclusion
- References
- Globally Optimal Image Partitioning by Multicuts
- Introduction
- Problem Description
- Image Labeling
- Unsupervised Pairwise Image Partitioning
- The Multicut Problem
- Image Labeling as Multicut Problem
- Finding an Optimal Multicut
- Linear Programming Formulations
- Rounding Fractional Solutions
- Finding Violated Constraints
- Experiments
- Conclusions
- References
- A Fast Solver for Truncated-Convex Priors: Quantized-Convex Split Moves
- Introduction
- Related Work
- Move Algorithms
- Quantized Move
- Convex Moves
- Truncated Convex Prior Algorithm
- Results
- Conclusion and Future Work
- References
- Continuous Optimization
- Temporally Consistent Gradient Domain Video Editing
- Introduction
- Video Editing Model (Continuous Setting)
- Discretization of the Model
- Discretization of the Convective Derivative
- The Deblurring Convective Derivative (DCD)
- Applications
- Conclusions and Future Work
- References
- Texture Segmentation via Non-local Non-parametric Active Contours
- Introduction
- Previous Works
- Comparison with Previous Works
- Contributions
- Non-local Active Contours
- Un-normalized Non-local Active Contours [10,11]
- Normalized Non-local Active Contour Model
- Wasserstein Local Homogeneity
- Wasserstein Distance
- Sliced Wasserstein Distance
- Wasserstein Non-local Active Contours
- Experimental Results and Comparisons
- Conclusion
- References
- Evaluation of a First-Order Primal-Dual Algorithm for MRF Energy Minimization
- Introduction
- Overview
- Related Work and Motivation
- Contribution
- Methods
- LP Relaxation of the MAP Problem
- Primal-Dual Iteration Scheme
- Estimating Primal and Dual Bounds
- Experimental Results
- Conclusion
- References
- Global Relabeling for Continuous Optimization in Binary Image Segmentation
- Introduction
- Preliminaries
- Binary Segmentation Models
- Graph Cuts
- Total Variation Formulation
- Primal-Dual Gap
- Algorithm
- Primal Dual Optimization
- Global Relabeling
- Experimental Results
- Conclusion
- References
- Stop Condition for Subgradient Minimization in Dual Relaxed (max,+) Problem
- Introduction: Definition of Main Concepts
- Equivalent Transformations of a Relaxed Labeling Problem into Trivial One
- Subgradient Descent
- Stop Condition of the Subgradient Descent Algorithm
- Finding the Upper and Lower Bound of the Problem's Slack
- Minimizing the Upper Bound
- Experiments
- Vertical and Horizontal Lines
- Segmentation
- Concluding Remarks
- References
- Segmentation
- Optimality Bounds for a Variational Relaxation of the Image Partitioning Problem
- Introduction
- Probabilistic Rounding and the Coarea Formula
- Probabilistic Rounding for Multiclass Image Partitions
- A Probabilistic A Priori Optimality Bound
- Experiments
- Conclusion
- References
- Interactive Segmentation with Super-Labels
- Introduction
- Related Work
- Modeling Complex Appearance via Super-Labels
- Problem Formulation
- Our SUPERLABELSEG Algorithm
- Applications and Experiments
- Binary Segmentation
- Complex Appearance Models
- Multi-Class Segmentation
- Interactive Co-segmentation
- Discussion: Super-Labels as Semi-supervised Learning
- Conclusion
- References
- Curvature Regularity for Multi-label Problems - Standard and Customized Linear Programming
- Introduction
- Curvature and Linear Programming
- Formulations for Multi-label Problems
- Multi-Region Segmentation
- Inpainting
- Denoising
- Alternative Strategies
- Linear Programming and Higher Order MRFs
- Do We Need Standard Solvers?
- Experiments
- Image Segmentation
- Inpainting
- Denoising
- Comparison of ILPs
- Conclusion
- References
- Space-Varying Color Distributions for Interactive Multiregion Segmentation: Discrete versus Continuous Approaches
- Introduction
- A Statistical Framework for Segmentation
- Segmentation as Bayesian Inference
- Inferring Space-Variant Color Distributions
- Discrete versus Continuous Energies
- MRF Approaches
- PDE Approaches
- Advantages and Disadvantages
- Comparison of Segmentation Accuracy
- Results on the Graz Benchmark
- Visual Results
- Ambiguities
- Metrication Errors
- Runtimes
- Conclusion
- References
- Detachable Object Detection with Efficient Model Selection
- Introduction
- Related Work and Key Idea
- Background: From Local Occlusion Ordering to Global Consistency
- Automatic Model Selection: Formulation
- Automatic Model Selection: Implementation
- Experiments
- Qualitative Performance
- Failure Modes
- Quantitative Assessment
- Discussion
- References
- Curvature Regularization for Curves and Surfaces in a Global Optimization Framework
- Curvature in Vision
- Length and Area Regularization
- Curvature Regularization
- Avoiding Extraneous Arcs
- Curvature of Surfaces
- Pseudo-Boolean Optimization
- Tessellations
- Hexagonal Meshes
- Adaptive Meshes
- Experimental Results
- Hexagonal Meshes
- Adaptive Meshes
- The Wilmore Functional
- Pseudo-Boolean Optimization
- Conclusions
- References
- SlimCuts: GraphCuts for High Resolution Images Using Graph Reduction
- Introduction
- Prior Work
- Contribution
- Segmentation by Discrete Energy Minimization
- Constructing Slim Graphs
- Slim Graphs for Simplified User Interaction
- Experiments
- Experiments on Small Scale Images
- Experiments on Large Scale Images
- Experiments on Resource-Limited Systems
- Conclusion
- References
- Discrete Optimization of the Multiphase Piecewise Constant Mumford-Shah Functional
- Introduction
- Previous Work
- Methods
- Review of Level Set Formulation of the Multiphase Mumford Shah model
- Discrete Formulation
- Discrete Optimization
- Algorithm
- Experimental Results
- Conclusion
- References
- Image Segmentation with a Shape Prior Based on Simplified Skeleton
- Introduction
- Related Work
- Contribution
- An Iterative Approach to Segmentation with a Shape Prior
- Iterative Segmentation as a Coordinate Descent
- Relation to EM Algorithm
- Simplified Figure Skeleton as a Shape Prior
- Graph-Based Shape Model
- Unary Potentials
- Shape Fitting via Simulated Annealing
- Experiments
- Unary and Pairwise Terms
- Coordinate Descent
- Giraffe Segmentation
- Letter Segmentation
- Conclusion
- References
- High Resolution Segmentation of Neuronal Tissues from Low Depth-Resolution EM Imagery
- Introduction
- Approach
- Learning a Discriminative, over-Complete Dictionary
- SVM Classifier
- Experiments
- Parameter Selection
- Simulations with FIB Data
- ssTEM Data
- Conclusion
- References
- Motion and Video
- Optical Flow Guided TV-L1 ideo Interpolation and Restoration
- Introduction
- The Optical Flow Guided TV-L1 Model
- Minimizing the Optical Flow Guided TV-L1 Model
- Discretization
- Primal-Dual Algorithm
- The Optical Flow Guided TV-L1 Model for Frame Interpolation
- Application and Numerical Evaluations
- Denoising
- Inpainting
- Image Sequence Interpolation
- Conclusion
- References
- Data-Driven Importance Distributions for Articulated Tracking
- Motivation
- Articulated Tracking Using Particle Filters
- Related Work
- A Failed Experiment
- Spatial Predictions
- Data-Driven Importance Distributions
- An Importance Distribution Based on Silhouettes
- An Importance Distribution Based on Depth
- A Simple Likelihood Model
- Experimental Results
- Conclusion
- References
- Robust Trajectory-Space TV-L1 Optical Flow for Non-rigid Sequences
- Introduction
- Related Work and Contribution
- Multi-frame Image Alignment
- Subspace Trajectory Model
- Choice of Basis
- Variational Multi-frame Optical Flow Estimation
- Optimization of the Proposed Energy
- Minimization of Step 1
- Minimization of Step 2
- Implementation Details
- Experimental results
- Construction of a Ground Truth Benchmark Dataset
- Quantitative Results on Benchmark Sequence
- Experiments on Real Sequences
- Conclusions
- References
- TV-L1 Optical Flow for Vector Valued Images
- Introduction
- TV-L1 Optical Flow of Vector Valued Images
- A General Minimization Problem
- Implementation
- Projections on Elliptic Balls
- Implementation Choices
- Examples
- Results
- Conclusion and Future Research
- References
- Using the Higher Order Singular Value Decomposition for Video Denoising
- Introduction
- Theory
- Implementation of the HOSVD for Video Denoising
- Choice of Patch Similarity Measure
- HOSVD and Universal 3D Transforms
- Experimental Results and Comparisons
- Conclusion
- References
- Learning
- Optimization of Robust Loss Functions for Weakly-Labeled Image Taxonomies: An Image Net Case Study
- Introduction
- Literature Review
- Problem Statement
- The Loss Function
- 'Boosting' of Binary Classifiers
- The Latent Setting
- The Optimization Problem
- Convex Relaxation
- Learning the Latent Variables
- Column Generation
- Experiments
- Binary Classifiers
- Structured Classifiers
- Conclusion
- References
- Multiple-Instance Learning with Structured Bag Models
- Introduction
- Structured Bag Models
- Model 1 (Hard Constraints)
- Model 2 (Soft Constraints)
- Model 3 (Combined Model)
- Inference Using Dual Decomposition
- Model 1 (Hard Constraints)
- Model 2 (Soft Constraints)
- Model 3 (Combined Model)
- Training Using Deterministic Annealing
- Model 1 (Hard Constraints)
- Model 2 (Soft Constraints)
- Model 3 (Combined Model)
- Semi-supervised Segmentation with Weak MIL Annotations (Road Crack Detection)
- MIL-Based Interactive Cosegmentation
- Discussion
- References
- Branch and Bound Strategies for Non-maximal Suppression in Object Detection
- Introduction
- Related Work
- The Energy
- Minimization of a Supermodular Function
- Branch and Bound Implementations
- Theoretical Results
- Empirical Results
- Discussion
- Conclusions
- References
- Shape Analysis
- Metrics, Connections, and Correspondence: The Setting for Groupwise Shape Analysis
- Introduction
- The Metric
- Global and Local Re-parameterisations
- Correspondence and Connections
- Optimising Correspondence
- Discussion
- References
- The Complex Wave Representation of Distance Transforms
- Introduction
- The Complex Wave Representation (CWR)
- Distance Transform Gradient Density
- Properties of the Fourier Transform of the CWR
- Spatial Frequencies as Gradient Histogram Bins
- Power Spectrum of ?(x, y) as a Gradient Density Estimator
- Empirical Confirmation of the Main Result
- Discussion
- References
- Author Index
