An Improved Algorithm for TV-L 1 Optical Flow

By Andreas Wedel, Thomas Pock, Christopher Zach, Horst Bischof, Daniel Cremers. Statistical and Geometrical Approaches to Visual Motion Analysis, 2009

510 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/0f671859d83f6c19e0af5f5ccfe8aad3681ccc1b

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

Multi-Modality Fusion based on Consensus-Voting and 3D Convolution for Isolated Gesture Recognition (2016, methodology): “The optical ﬂow ﬁelds are acquired using [40].”
PWC-Net: CNNs for Optical Flow Using Pyramid, Warping, and Cost Volume (2017, methodology): “Finally, a common practice by the traditional methods is to post-process the optical flow using contextual information, such as median filtering [49] and bilateral filtering [54].”
Dense Point Trajectories by GPU-Accelerated Large Displacement Optical Flow (2010, methodology): “Very efficient GPU implementations of other variational optical flow models have been proposed in [3, 13 ,14].”
Live dense reconstruction with a single moving camera (2010, background): “[17] also includes photometric normalisation, improving robustness to varying lighting conditions or automatic camera parameters.”
A Database and Evaluation Methodology for Optical Flow (2007, methodology): “Huber-L1 [82] 2 DPOF [39] 261 TV-L1-improved [80] 2.”
Optical flow modeling and computation: A survey (2015, methodology): “Another way to obtain robustness against illumination changes is to work with the cartoon and texture components of the image, as initiated in [263].”
Motion estimation via hierarchical block matching and graph cut (2015, methodology): “EXPERIMENTS In all the experiments, we apply the Rudin-Osher-Fatemi (ROF) structure texture decomposition method [13] to preprocess the input sequences to alleviate the impact of lighting change.”
A primal-dual framework for real-time dense RGB-D scene flow (2015, background): “The L1 norm has shown to be more robust against outliers than the L2 norm [18] and the choice of the primal-dual algorithm to solve the minimization problem allows us to minimize it exactly without resorting to any approximation (more details will be given in section III).”
Stereoscopic Scene Flow Computation for 3D Motion Understanding (2011, methodology): “For a more detailed explanation of the algorithm we refer to Wedel et al. (2009).”
A Unified Framework for Multi-Modal Isolated Gesture Recognition (2018, methodology): “The optical ﬂow ﬁelds are acquired using [Wedel et al. 2009].”
Weakly Supervised Action Localization by Sparse Temporal Pooling Network (2017, methodology): “For the ﬂow stream, we apply the TV-L 1 optical ﬂow algorithm [39] and truncate the ﬂow magnitude to be in the range of [ − 20 , 20] .”
3D scene flow estimation with a rigid motion prior (2011, methodology): “To alleviate the negative influence of varying illumination on the data term, we preprocess the input images using structure-texture decomposition [2], which has been used successfully for 2D optical flow estimation – for details see [21].”
Layered image motion with explicit occlusions, temporal consistency, and depth ordering (2010, methodology): “The method achieves state-of-the-art results on the Middlebury benchmark and produces meaningful scene segmentations as well as detected occlusion regions.”
Ieee Transactions on Pattern Analysis and Machine Intelligence 1 Large Displacement Optical Flow: Descriptor Matching in Variational Motion Estimation (2011, unknown): “An alternative feature with similar effects can be obtained by structure-texture decomposition [34].”
TV-L1 Optical Flow Estimation (2013, background): “It also affects the running time, so it is a compromise between speed and accuracy.”
Simultaneous Optical Flow and Intensity Estimation from an Event Camera (2016, methodology): “These are, in essence, smoothness priors which have been applied in many image processing application for standard cameras, such as optical flow [23], image denoising [21] and SLAM [16].”