CASIA OpenIR  > 机器人视觉团队
High-quality 3D Reconstruction with Depth Super-resolution and Completion
Li JW(李建伟)1,2; Gao W(高伟)1,2; Wu YH(吴毅红)1,2
Source PublicationIEEE Access
2019
Volume7Issue:1Pages:19370-19381
Abstract
3D reconstruction is an important topic in computer vision with many applications, such as robotics and augmented reality. Since the raw depth images captured by consumer RGB-D cameras are often low-resolution, noisy, and incomplete. How to obtain high-quality 3D models with a consumer RGB-D camera is still a challenge for existing systems. In this paper, we propose a new depth super-resolution and completion (SRC) method implemented in deep learning framework, and build a high-quality 3D reconstruction system. We first improve the resolution of low-resolution (LR) depth image with a depth super-resolution network, and remove the outliers in high-resolution (HR) depth image based on gradient saliency. To further enhance the quality of HR depth image with the guide of HR color image, we learn surface normal and occlusion boundary images from the corresponding HR color image through two deep fully convolutional networks (FCN). In particular, the blurriness of HR color image is also detected and pixel-wise quantized. Finally, we obtain a completed HR depth image by optimizing the HR depth image with surface normal, occlusion boundary and color image blurriness. We have carried out qualitative and quantitative evaluations with baseline methods on public datasets. Experimental results demonstrate that our method has better performance both on single depth image enhancement and 3D reconstruction.
KeywordDeep Learning Super-resolution Image Processing Depth Completion 3d Reconstruction
Indexed BySCIE
Language英语
WOS IDWOS:000459939900001
Citation statistics
Document Type期刊论文
Identifierhttp://ir.ia.ac.cn/handle/173211/23361
Collection机器人视觉团队
Corresponding AuthorGao W(高伟)
Affiliation1.中科院自动化研究所
2.中国科学院大学
Recommended Citation
GB/T 7714
Li JW,Gao W,Wu YH. High-quality 3D Reconstruction with Depth Super-resolution and Completion[J]. IEEE Access,2019,7(1):19370-19381.
APA Li JW,Gao W,&Wu YH.(2019).High-quality 3D Reconstruction with Depth Super-resolution and Completion.IEEE Access,7(1),19370-19381.
MLA Li JW,et al."High-quality 3D Reconstruction with Depth Super-resolution and Completion".IEEE Access 7.1(2019):19370-19381.
Files in This Item: Download All
File Name/Size DocType Version Access License
FINAL Article.pdf(4152KB)期刊论文作者接受稿开放获取CC BY-NC-SAView Download
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[Li JW(李建伟)]'s Articles
[Gao W(高伟)]'s Articles
[Wu YH(吴毅红)]'s Articles
Baidu academic
Similar articles in Baidu academic
[Li JW(李建伟)]'s Articles
[Gao W(高伟)]'s Articles
[Wu YH(吴毅红)]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[Li JW(李建伟)]'s Articles
[Gao W(高伟)]'s Articles
[Wu YH(吴毅红)]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: FINAL Article.pdf
Format: Adobe PDF
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.