A Taxonomy and Evaluation of Dense Two-Frame Stereo Correspondence Algorithms

By Daniel Scharstein, Richard Szeliski. 2001

8,091 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Ocean Engineering, Evaluation & Benchmarks, 3D Vision & Pose

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

In Ocean Engineering

In Evaluation & Benchmarks

In 3D Vision & Pose

Citations in the corpus

CIDEr: Consensus-based image description evaluation (2014, background): “This has already been shown in various problems in computer vision, such as detection [13, 7], segmentation [13, 28], and stereo [39].”
IRS: A Large Synthetic Indoor Robotics Stereo Dataset for Disparity and Surface Normal Estimation (2019, background): “MiddleBury is a well-known and frequently updated real world stereo dataset [23], [10], [11], [24], [12], which takes the resolution, brightness, exposure and many other uncommon factors into consideration to improve the complexities of the dataset and provides dense per-pixel disparities of the sce”
Learning a Multi-View Stereo Machine (2017, methodology): “We refer the reader to [47] for an overview of classical binocular stereo matching algorithms.”
Depth Map Prediction from a Single Image using a Multi-Scale Deep Network (2014, methodology): “While there is much prior work on estimating depth based on stereo images or motion [17], there has been relatively little on estimating depth from a single image.”
A Large Dataset to Train Convolutional Networks for Disparity, Optical Flow, and Scene Flow Estimation (2015, background): “The first significant efforts to create standard datasets were the Middlebury datasets for stereo disparity estimation [20] and optical flow estimation [1].”
An experimental comparison of min-cut/max- flow algorithms for energy minimization in vision (2001, background): “Index Terms — Energy minimization, graph algorithms, minimum cut, maximum flow, image restoration, segmentation, stereo, multi-camera scene reconstruction.”
Playing for Benchmarks (2017, background): “The Middlebury benchmarks for stereo, multi-view stereo, and optical ﬂow played a pivotal role in the development of low-level vision algorithms [4, 57, 58, 59].”
SIFT Flow: Dense Correspondence across Scenes and Its Applications (2011, methodology): “Moreover, we want to have a simple, effective, object-free model to align image pairs such as the ones in Figure 1 (c).”
MOT16: A Benchmark for Multi-Object Tracking (2016, background): “…past, the computer vision community has developed centralized benchmarks for numerous tasks including object detection [14], pedestrian detection [12], 3D reconstruction [39], optical ﬂow [7], [19], visual odometry [19], single-object short-term tracking [26], and stereo estimation [19], [37].”
Stereo Matching by Training a Convolutional Neural Network to Compare Image Patches (2015, methodology): “According to the taxonomy of Scharstein and Szeliski (2002), a typical stereo algorithm consists of four steps: matching cost computation, cost aggregation, optimization, and disparity reﬁnement.”
Pyramid Stereo Matching Network (2018, methodology): “According to [25], a typical stereo matching algorithm consists of four steps: matching cost computation, cost aggregation, optimization, and disparity refinement.”
Autonomous flight in unstructured and unknown indoor environments (2009, background): “However, more recently, researchers have used global image information t o i fer the depth of pixels for which there is no unambiguous correspondence between image s [83, 30, 87].”
Evaluation of Cost Functions for Stereo Matching (2007, methodology): “More complicated similarity measures are possible, including mutual information [7, 9, 11] and approximative Recent stereo surveys [5, 17] and the Middlebury online evaluation [14] compare state-of-the-art stereo methods on test data with complex geometries and varied texture.”
Fast Edge-Preserving PatchMatch for Large Displacement Optical Flow (2014, neutral): “Our method follows the traditional local correspondence searching framework [20] which consists of 4 steps:”
HITNet: Hierarchical Iterative Tile Refinement Network for Real-time Stereo Matching (2020, methodology): “Stereo matching has been an active field of research for decades [36, 43, 19].”