A Quantitative Evaluation of Confidence Measures for Stereo Vision

By Xiaoyan Hu, Philippos Mordohai. 2012

270 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

Citations in the corpus

• Robust Probabilistic Occupancy Grid Estimation from Positive and Negative Distance Fields (2012, unknown): “Based on a recent comparison of confidence measures [15], we selected a form of confidence introduced”
• Multi-View Stereo: A Tutorial (2015, unknown): “In addition to the depth value with the highest photo-consistency, the algorithm often evaluates a confidence measure so that low-confidence depth values can be ignored or down-weighted in the later model merging step [106].”
• High-Resolution Stereo Datasets with Subpixel-Accurate Ground Truth (2014, unknown): “Stereo vision is one of the most heavily researched topics in computer vision [5, 17, 18, 20, 28], and much of the progress over the last decade has been driven by the availability of standard test images and benchmarks [7, 14, 27, 28, 30, 31].”
• Intel RealSense Stereoscopic Depth Cameras (2017, unknown): “ators [31] is implemented, similar to those used in standard matching literature [10].”
• LAF-Net: Locally Adaptive Fusion Networks for Stereo Confidence Estimation (2019, unknown): “From matching cost, the peak ratio of the matching costs [11] and naive peak ratio [12] have been widely used to remove unreliable pixels.”
• Ensemble Learning for Confidence Measures in Stereo Vision (2013, unknown): “In the absense of a strong theoretical foundation to account for properties of global energies in commonplace stereo aggregation schemes, many spatially local stereo confidence measures have been proposed [5] [13] .”
• Robust stereo matching with surface normal prediction (2017, unknown): “So the depth prior constraint can be defined as Eobs = ||W(z− z0)||22, (12) where z denotes the variable vector of disparity, z0 denotes the vector of pre-computed disparity, and W is a diagonal matrix indicating weights, which is defined as W = diag(w1, w2, . . . , wM ), (13) wi = ⎧⎨ ⎩ CP (xi, yi) ”
• Real-Time Self-Adaptive Deep Stereo (2018, unknown): “Batsos et al [2] soften this effect by combining traditional functions and confidence measures [15, 31] within a random forest framework, proving better generalization compared to CNN-based method [44].”
• Towards Probabilistic Volumetric Reconstruction Using Ray Potentials (2015, unknown): “ods assign confidence scores for each reconstructed 3D point [11, 15] or model the ambiguity in view-centric depth maps [9, 25, 31], we focus on volumetric reconstruction in this paper.”
• Learning to Detect Ground Control Points for Improving the Accuracy of Stereo Matching (2014, unknown): “The rationale here is that a large difference may indicate an unambiguous disparity assignment.”
• On the Uncertainty of Self-Supervised Monocular Depth Estimation (2020, unknown): “In these latter cases, estimating the uncertainty (or, complementary, the confidence) proved to be effective for depth-from-stereo, by means of both modelbased [24] and learning-based [55, 30] methods, optical flow [27], and semantic segmentation [26, 30].”
• A Survey on Deep Learning Techniques for Stereo-Based Depth Estimation (2020, unknown): “Conventional confidence estimation methods are mostly based on assumptions and heuristics on the matching cost volume analysis, see [59] for a review and evaluation of the early methods.”