摘 要:通过对双目视觉原理及亚像素插值技术的深入分析,研究如何在双目视觉图像的基础上实现场景的高精度3D 重建。实验以Intel RealSense D435 相机作为采集工具,完成摄像机标定和图像预处理,相机拍摄的图像为本次研究提供了较为清晰的数据展现,利用OpenCV 和SIFT 算法完成立体匹配,获取点云图。从实验结果来看,该技术能够较好地实现双目视觉条件下的3D 重建。
关键词:三维重建;点云图;OpenCV;SIFT 算法
DOI:10.19850/j.cnki.2096-4706.2021.21.024
中图分类号:TP391.4 文献标识码:A 文章编号:2096-4706(2021)21-0094-04
High Precision 3D Reconstruction Technology Based on Binocular Vision
CAI Guihang, CHEN Qiang, WU Linfeng
(School of Electronic Information Engineering, Foshan University, Foshan 528225, China)
Abstract: Through the in-depth analysis of binocular vision principle and sub-pixel interpolation technology, this paper studies how to realize high-precision 3D reconstruction of scene based on binocular vision image. The experiment takes Intel RealSense D435 camera as the acquisition tool, completes camera calibration and image preprocessing, the images taken by the camera provide a clearer data display for this study, uses OpenCV and SIFT algorithm to complete stereo matching and obtain point cloud map. The experimental results show this technology can better realize 3D reconstruction under the condition of binocular vision.
Keywords: 3D reconstruction; point cloud map; OpenCV; SIFT algorithm
参考文献:
[1] 谭健胜.基于双目立体视觉的三维重建技术研究及实现[D]. 广州:华南理工大学,2017.
[2] 李杰. 基于双目视觉的深度重建技术的研究和应用 [D].长春:吉林大学,2016.
[3] 李彪. 基于双目立体视觉三维重建技术研究 [D]. 西安:西安科技大学,2018
[4] 吉玲. 双目立体视觉中三维重建关键技术研究 [D]. 天津:天津理工大学,2015.[ 5 ] WANG Q Y,WANG Z Y,SHANG Z D. Parameter calibration of a vision sensor with a geometric similarity constraint [J].Measurement Science and Technology,2020,31(3):1-8.
[6] 曾文献,郭兆坤. 立体匹配算法研究综述 [J]. 河北省科学院学报,2018,35(2):43-48.
[7] YOO J C,HAN T H. Fast normalized cross-correlation [J].Circuits Systems and Signal Processing,2009,28(6):819-843.
作者简介:蔡桂航(2000—),男,汉族,广东佛山人,本科在读,研究方向:计算机视觉。