Simultaneous Stereo Matching and Confidence Estimation Network

Author:

Schmähling Tobias1ORCID,Müller Tobias1ORCID,Eberhardt Jörg1,Elser Stefan2

Affiliation:

1. Institute for Photonic Systems Hochschule Ravensburg-Weingarten, University of Applied Sciences, Doggenriedstraße, 88250 Weingarten, Germany

2. Institute for Artificial Intelligence Hochschule Ravensburg-Weingarten, University of Applied Sciences, Doggenriedstraße, 88250 Weingarten, Germany

Abstract

In this paper, we present a multi-task model that predicts disparities and confidence levels in deep stereo matching simultaneously. We do this by combining its successful model for each separate task and obtaining a multi-task model that can be trained with a proposed loss function. We show the advantages of this model compared to training and predicting disparity and confidence sequentially. This method enables an improvement of 15% to 30% in the area under the curve (AUC) metric when trained in parallel rather than sequentially. In addition, the effect of weighting the components in the loss function on the stereo and confidence performance is investigated. By improving the confidence estimate, the practicality of stereo estimators for creating distance images is increased.

Funder

Federal Ministry of Education and Research

Publisher

MDPI AG

Reference51 articles.

1. A taxonomy and evaluation of dense two-frame stereo correspondence algorithms;Scharstein;Int. J. Comput. Vis.,2002

2. Hartley, R., and Zisserman, A. (2003). Multiple View Geometry in Computer Vision, Cambridge University Press.

3. Hannah, M.J. (1974). Computer Matching of Areas in Stereo Images, Stanford University.

4. PMF: A stereo correspondence algorithm using a disparity gradient limit;Pollard;Perception,1985

5. Real-time stereo and motion integration for navigation;Baker;Proceedings of the ISPRS Commission III Symposium: Spatial Information from Digital Photogrammetry and Computer Vision,1994

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3