On building an accurate stereo matching system on graphics hardware

By Xing Mei, Xun Sun, Mingcai Zhou, Shaohui Jiao, Haitao Wang, Xiaopeng Zhang. IEEE International Conference on Computer Vision, 2011

612 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/0aa180ec435b3168f23982580eb1c4aa6e815072

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

LAF-Net: Locally Adaptive Fusion Networks for Stereo Confidence Estimation (2019, background): “However, because of its challenging elements such as reﬂective surfaces, textureless regions, repeated pattern regions, occlusions [23, 13, 6], and photometric deformations incurred by illumination and camera speciﬁca-tion variations [44, 9], the stereo matching still remains an unsolved problem.”
Intel RealSense Stereoscopic Depth Cameras (2017, methodology): “Additionally, the correlation cost function has been expanded beyond simple Census correlation, integrating other matching measures [17].”
Multi-View Stereo: A Tutorial (2015, methodology): “As a result, some techniques [141] combine SAD with another measure such as NCC or Census.”
Displets: Resolving stereo ambiguities using object knowledge (2015, background): “Thus, recent efforts mainly focus on global methods [2, 16, 40, 41, 54, 60] which are able to overcome some of these problems by imposing smoothness constraints between adjacent pixels or superpixels.”
Fast Edge-Preserving PatchMatch for Large Displacement Optical Flow (2014, positive): “In our implementation, we use the AD-Census [40] for computing matching cost (i.e., the CIELab color cue together with the census transform cue).”
Stereo Matching by Training a Convolutional Neural Network to Compare Image Patches (2015, methodology): “The stereo method we used was inﬂuenced by Mei et al. (2011) and comprises cross-based cost aggregation, semiglobal matching, a left-right consistency check, subpixel enhancement, a median, and a bilateral ﬁlter.”
High-Accuracy Stereo Matching Based on Adaptive Ground Control Points (2015, background): “Mei [33] get the best results when the error threshold is 1.0 and 2.0.”
Computing the stereo matching cost with a convolutional neural network (2014, methodology): “Peris et al. [12] initialized the matching cost with AD-Census [11] and used multiclass linear discriminant analysis to learn a mapping from the computed matching cost to the final disparity.”