Saved in:
Bibliographic Details
Main Authors: Feng, Xiang, Kang, Kaizhang, Pei, Fan, Ding, Huakeng, You, Jinjiang, Tan, Ping, Zhou, Kun, Wu, Hongzhi
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2308.03492
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910738142986240
author Feng, Xiang
Kang, Kaizhang
Pei, Fan
Ding, Huakeng
You, Jinjiang
Tan, Ping
Zhou, Kun
Wu, Hongzhi
author_facet Feng, Xiang
Kang, Kaizhang
Pei, Fan
Ding, Huakeng
You, Jinjiang
Tan, Ping
Zhou, Kun
Wu, Hongzhi
contents We propose a novel framework to automatically learn to aggregate and transform photometric measurements from multiple unstructured views into spatially distinctive and view-invariant low-level features, which are subsequently fed to a multi-view stereo pipeline to enhance 3D reconstruction. The illumination conditions during acquisition and the feature transform are jointly trained on a large amount of synthetic data. We further build a system to reconstruct both the geometry and anisotropic reflectance of a variety of challenging objects from hand-held scans. The effectiveness of the system is demonstrated with a lightweight prototype, consisting of a camera and an array of LEDs, as well as an off-the-shelf tablet. Our results are validated against reconstructions from a professional 3D scanner and photographs, and compare favorably with state-of-the-art techniques.
format Preprint
id arxiv_https___arxiv_org_abs_2308_03492
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Learning Photometric Feature Transform for Free-form Object Scan
Feng, Xiang
Kang, Kaizhang
Pei, Fan
Ding, Huakeng
You, Jinjiang
Tan, Ping
Zhou, Kun
Wu, Hongzhi
Computer Vision and Pattern Recognition
We propose a novel framework to automatically learn to aggregate and transform photometric measurements from multiple unstructured views into spatially distinctive and view-invariant low-level features, which are subsequently fed to a multi-view stereo pipeline to enhance 3D reconstruction. The illumination conditions during acquisition and the feature transform are jointly trained on a large amount of synthetic data. We further build a system to reconstruct both the geometry and anisotropic reflectance of a variety of challenging objects from hand-held scans. The effectiveness of the system is demonstrated with a lightweight prototype, consisting of a camera and an array of LEDs, as well as an off-the-shelf tablet. Our results are validated against reconstructions from a professional 3D scanner and photographs, and compare favorably with state-of-the-art techniques.
title Learning Photometric Feature Transform for Free-form Object Scan
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2308.03492