High accuracy and geometry-consistent confidence prediction network for multi-view stereo

doi:10.1016/j.cag.2021.04.020

	High accuracy and geometry-consistent confidence prediction network for multi-view stereo
	Li, Zhaoxin 1; Zhang, Xiaoge 1; Wang, Kangkan 2; Jiang, Hao 1; Wang, Zhaoqi 1
	2021-06-01
发表期刊	COMPUTERS & GRAPHICS-UK
ISSN	0097-8493
卷号	97 页码:148-159
摘要	Confidence prediction task attempts to infer the correctness of estimated depth hypotheseshich has gained popularity recently in stereo matching and boosts the accuracy of disparity estimation. However, less attention is paid on confidence prediction of multi-view stereo (MVS), where multi-view depth estimation is a key step for high-quality reconstruction. In this work, we propose a Geometry-consistent Confidence prediction Network (GeoConfNet), where the correctness of a depth hypothesis is accurately predicted via a deep neural network that explores both spatial coherence and cross-view consistency. The proposed deep network consists of a feature extraction module, a U-Net-based fusion module and a confidence refinement module. Furthermore, we demonstrate that truncated signed distance field (TSDF) is a powerful cross-view feature which can be an effective complement to spatial features, thereby remarkably boosting confidence prediction accuracy of MVS. Exhaustive experiments on a variety of MVS datasets as well as stereo matching datasets clearly demonstrate that our method achieves significantly better performance than state-of-the-art methods in terms of area under the curve (AUC). (c) 2021 Elsevier Ltd. All rights reserved.
关键词	3D reconstruction Confidence prediction Multi-view stereo PatchMatch stereo
DOI	10.1016/j.cag.2021.04.020
收录类别	SCI
语种	英语
资助项目	National Key Research and Development Program of China[2018AAA0103002] ; National Key Research and Development Program of China[2017YFB1002600] ; National Natural Science Foundation of China[61702482]
WOS研究方向	Computer Science
WOS类目	Computer Science, Software Engineering
WOS记录号	WOS:000661427000003
出版者	PERGAMON-ELSEVIER SCIENCE LTD
引用统计	被引频次：4[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/17652
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Li, Zhaoxin; Jiang, Hao
作者单位	1.Chinese Acad Sci, Inst Comp Technol, Beijing, Peoples R China 2.Nanjing Univ Sci & Technol, Sch Comp Sci & Engn, Nanjing, Peoples R China
推荐引用方式 GB/T 7714	Li, Zhaoxin,Zhang, Xiaoge,Wang, Kangkan,et al. High accuracy and geometry-consistent confidence prediction network for multi-view stereo[J]. COMPUTERS & GRAPHICS-UK,2021,97:148-159.
APA	Li, Zhaoxin,Zhang, Xiaoge,Wang, Kangkan,Jiang, Hao,&Wang, Zhaoqi.(2021).High accuracy and geometry-consistent confidence prediction network for multi-view stereo.COMPUTERS & GRAPHICS-UK,97,148-159.
MLA	Li, Zhaoxin,et al."High accuracy and geometry-consistent confidence prediction network for multi-view stereo".COMPUTERS & GRAPHICS-UK 97(2021):148-159.