Custom cover image
Stereo Scene Flow for 3D Motion Analysis / by Andreas Wedel, Daniel Cremers
Contributor(s): Resource type: Ressourcentyp: Buch (Online)Book (Online)Language: English Series: SpringerLink BücherPublisher: London : Springer-Verlag London Limited, 2011Description: Online-Ressource (IX, 128p. 74 illus., 60 illus. in color, digital)ISBN:- 9780857299659
- 006.6
- 006.37
- TA1637-1638 TA1637-1638
- TA1630
Contents:
Summary: Daniel CremersSummary: This book presents methods for estimating optical flow and scene flow motion with high accuracy, focusing on the practical application of these methods in camera-based driver assistance systems. Clearly and logically structured, the book builds from basic themes to more advanced concepts, culminating in the development of a novel, accurate and robust optic flow method. Features: reviews the major advances in motion estimation and motion analysis, and the latest progress of dense optical flow algorithms; investigates the use of residual images for optical flow; examines methods for deriving motPPN: PPN: 1651017646Package identifier: Produktsigel: ZDB-2-SCS
Stereo Scene Flow for 3D Motion Analysis; Preface; Contents; List of Notations; Chapter 1: Machine Vision Systems; Chapter 2: Optical Flow Estimation; 2.1 Optical Flow and Optical Aperture; 2.2 Feature-Based Optical Flow Approaches; 2.2.1 Census Based Optical Flow; 2.2.2 The Optical Flow Constraint; 2.2.3 Lucas-Kanade Method; 2.3 Variational Methods; 2.3.1 Total Variation Optical Flow; 2.3.2 Quadratic Relaxation; 2.3.3 Large Displacement Flow: Novel Algorithmic Approaches; 2.3.4 Other Optical Flow Approaches; 2.4 The Flow Refinement Framework; 2.4.1 Data Term Optimization
2.4.1.1 Approximating the Absolute Function2.4.1.2 Quadratic Optimization; 2.4.1.3 Single Data Term; 2.4.1.4 Two Data Terms; 2.4.1.5 Multiple Data Terms; 2.4.1.6 Toy Example: Optical Flow Data Term; 2.4.2 Smoothness Term Evaluation; 2.4.2.1 Median Filtering; 2.4.2.2 TV-L1 Denoising; 2.4.2.3 TV-L2 Denoising; 2.4.2.4 Structure-Adaptive Smoothing; 2.4.2.5 Advanced Priors for Motion Estimation; 2.4.2.6 Toy Example Cont.: Denoising the Flow Field; 2.4.3 Implementation Details; 2.4.3.1 Pyramid Restriction and Prolongation
2.4.3.2 Re-sampling the Coefficients of the Optical Flow Data Term via Warping2.4.3.3 Toy Example Cont.: Pyramids and Warping; 2.4.3.4 Symmetric Gradients for the Data Term Evaluation; 2.4.3.5 Numerical Scheme; Chapter 3: Residual Images and Optical Flow Results; 3.1 Increasing Robustness to Illumination Changes; 3.2 Quantitative Evaluation of the Refinement Optical Flow; 3.2.1 Performance; 3.2.2 Smoothness Filters; TV-L2 (0.348 px avg. EPE, 2.17 sec run time); Felsberg Denoising (0.378 px avg. EPE, 5.94 sec run time); 2nd Order Smoothing (0.364 px avg. EPE, 61.5 sec run time)
TV-L1 (0.413 px avg. EPE, 1.0 sec run time using lambda=0.25 lambda=1.0 yields an avg. EPE of 0.429 px); Median Filtering (0.494 px avg. EPE, 3.21 sec run time for 20 iterations; number of iterations 2/20/200 yields avg. EPE of 0.553/0.494/0.533 in px); 3.2.3 Accuracy; 3.3 Results for Traffic Scenes; 3.4 Conclusion; Chapter 4: Scene Flow; 4.1 Visual Kinesthesia; 4.1.1 Related Work; 4.1.2 A Decoupled Approach for Scene Flow; 4.2 Formulation and Solving of the Constraint Equations; 4.2.1 Stereo Computation; 4.2.2 Scene Flow Motion Constraints; Occlusion Handling
4.2.3 Solving the Scene Flow Equations4.2.4 Evaluation with Different Stereo Inputs; 4.3 From Image Scene Flow to 3D World Scene Flow; Chapter 5: Motion Metrics for Scene Flow; 5.1 Ground Truth vs. Reality; 5.2 Derivation of a Pixel-Wise Accuracy Measure; 5.2.1 A Quality Measure for the Disparity; 5.2.2 A Quality Measure for the Scene Flow; 5.2.3 Estimating Scene Flow Standard Deviations; 5.3 Residual Motion Likelihood; 5.4 Speed Likelihood; Chapter 6: Extensions of Scene Flow; 6.1 Flow Cut-Moving Object Segmentation; Algorithm Overview; 6.1.1 Segmentation Algorithm; 6.1.1.1 Energy Functional
6.1.1.2 Graph Mapping
No physical items for this record