Detecting Ground Control Points via Convolutional Neural Network for Stereo Matching

TL;DR

This paper introduces a CNN-based method for detecting ground control points in stereo matching, which enhances disparity map accuracy by refining matching costs and optimizing with semi-global matching.

Contribution

It proposes a novel CNN-based GCP detection scheme and a GCP selection method that improves stereo matching accuracy.

Findings

Significant improvement in disparity map accuracy on KITTI 2012 dataset

Effective GCP detection using CNN trained on large stereo datasets

Enhanced matching cost refinement leads to better stereo matching results

Abstract

In this paper, we present a novel approach to detect ground control points (GCPs) for stereo matching problem. First of all, we train a convolutional neural network (CNN) on a large stereo set, and compute the matching confidence of each pixel by using the trained CNN model. Secondly, we present a ground control points selection scheme according to the maximum matching confidence of each pixel. Finally, the selected GCPs are used to refine the matching costs, and we apply the new matching costs to perform optimization with semi-global matching algorithm for improving the final disparity maps. We evaluate our approach on the KITTI 2012 stereo benchmark dataset. Our experiments show that the proposed approach significantly improves the accuracy of disparity maps.