Fast cost-volume filtering for visual correspondence and beyond

By Christoph Rhemann, Asmaa Hosni, Michael Bleyer, Carsten Rother, Margrit Gelautz. 2011

1,184 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Aerospace Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

Deep Stereo: Learning to Predict New Views from the World's Imagery (2015, methodology): “Figure 8: Comparison of real to synthesized KITTI images, showing results from the Kitti- and Street-View-trained model, a depth splatting algorithm from computed depth maps [29] and an optical flow algorithm [26].”
Unifying Flow, Stereo and Depth Estimation (2022, methodology): “Again, early optimization-based methods [55], [56] have been replaced by modern deep learning-based approaches [19], [24], [25], [57], [58], [59], [60], [61].”
Continuous Markov Random Fields for Robust Stereo Estimation (2012, neutral): “18 % CostFilter [38] 28.”
DTAM: Dense tracking and mapping in real-time (2011, background): “the cost volume (called a disparity space image in stereo matching [14], and generalised more recently in [10] for any discrete per pixel labelling) stores the accumulated photometric error as a function of inverse depth d.”
Are we ready for autonomous driving? The KITTI vision benchmark suite (2012, neutral): “06 % CostFilter [38] 19.”
Hierarchical Deep Stereo Matching on High-Resolution Images (2019, methodology): “To ensure global ordering and consistency constraints, global optimization techniques are applied with a considerable cost of time and memory, which also poses limitations on scale [10, 14].”
AANet: Adaptive Aggregation Network for Efficient Stereo Matching (2020, methodology): “Local stereo methods (either traditional [40, 12] or 2D/3D convolution based methods [20, 14]) usually perform window based cost aggregation:”
On building an accurate stereo matching system on graphics hardware (2011, methodology): “[9] formulated the aggregation step as a cost filtering problem.”
A hand pose tracking benchmark from stereo matching (2017, methodology): “There are a lot of stereo matching methods [11–14].”
GMFlow: Learning Optical Flow via Global Matching (2021, methodology): “, correlation) [13] was shown to be an effective component and thus has been extensively used in popular frameworks [16, 36, 39].”
StereoNet: Guided Hierarchical Refinement for Real-Time Edge-Aware Depth Prediction (2018, methodology): “Interestingly adaptive support weight approaches were cast as cost volume ﬁltering in [25]: a three-dimensional cost volume is constructed by computing the per-pixel matching costs at all possible disparity levels.”
PWC-Net: CNNs for Optical Flow Using Pyramid, Warping, and Cost Volume (2017, methodology): “Next, we use the features to construct a cost volume that stores the matching costs for associating a pixel with its corresponding pixels at the next frame [20].”
Dense Reconstruction Using 3D Object Shape Priors (2013, background): “[21]) [15] assumes known camera intrinsic and extrinsics and uses photo-consistency as a proxy for depth; viz, a point on a lambertian surface in the scene should project to the same colour in all the frames of the sequence, so the search for depth can be treated as a search for best photo-consisten”
Holoportation: Virtual 3D Teleportation in Real-time (2016, background): “Depth from RGB stereo is another possible solution that has seen several advances in real-time depth estimation [51, 61, 56, 52].”