Semi-dense Stereo Matching using Dual CNNs

TL;DR

This paper introduces a semi-dense stereo matching method using dual CNNs that incorporates global information and non-parametric transforms, achieving superior performance on standard datasets.

Contribution

It presents a novel dual CNN framework with global information integration and self-reliant transforms for improved semi-dense stereo matching.

Findings

Outperforms state-of-the-art semi-dense stereo methods on Middlebury dataset

Handles challenging cases like lighting changes and low textures effectively

Uses non-parametric transforms for parameter independence

Abstract

A robust solution for semi-dense stereo matching is presented. It utilizes two CNN models for computing stereo matching cost and performing confidence-based filtering, respectively. Compared to existing CNNs-based matching cost generation approaches, our method feeds additional global information into the network so that the learned model can better handle challenging cases, such as lighting changes and lack of textures. Through utilizing non-parametric transforms, our method is also more self-reliant than most existing semi-dense stereo approaches, which rely highly on the adjustment of parameters. The experimental results based on Middlebury Stereo dataset demonstrate that the proposed approach outperforms the state-of-the-art semi-dense stereo approaches.