CBMV: A Coalesced Bidirectional Matching Volume for Disparity Estimation

TL;DR

This paper introduces CBMV, a novel disparity estimation method that combines data-driven learning with traditional heuristics through a bidirectional matching volume and random forest classifiers, improving generalization.

Contribution

It presents a coalesced matching volume approach that integrates diverse evidence from bidirectional matching, enhancing generalization over purely data-driven methods.

Findings

Achieves similar accuracy to data-driven methods on benchmarks

Generalizes better to unseen data, including KITTI and ETH3D

Uses a classifier trained on Middlebury 2014 dataset

Abstract

Recently, there has been a paradigm shift in stereo matching with learning-based methods achieving the best results on all popular benchmarks. The success of these methods is due to the availability of training data with ground truth; training learning-based systems on these datasets has allowed them to surpass the accuracy of conventional approaches based on heuristics and assumptions. Many of these assumptions, however, had been validated extensively and hold for the majority of possible inputs. In this paper, we generate a matching volume leveraging both data with ground truth and conventional wisdom. We accomplish this by coalescing diverse evidence from a bidirectional matching process via random forest classifiers. We show that the resulting matching volume estimation method achieves similar accuracy to purely data-driven alternatives on benchmarks and that it generalizes to unseen data much better. In fact, the results we submitted to the KITTI and ETH3D benchmarks were generated using a classifier trained on the Middlebury 2014 dataset.