不完全角度背景纹影层析测量综述

1.北京航空航天大学 宁波创新研究院,浙江 宁波 315800; 2.北京航空航天大学 宇航学院,北京 100191

背景纹影; 层析技术; 三维重构; 不完全角度; 欠定病态方程

Review of incomplete-angle background-oriented schlieren tomography
HU Wei1,2,LI Jingxuan1,2,YANG Lijun1,2,ZHANG Yue2,LIANG Xuanye2

1.Ningbo Institute of Technology, Beihang University, Ningbo 315800, China; 2.School of Astronautics, Beihang University, Beijing 100191, China

background-oriented schlieren; tomographic technique; three-dimensional reconstruction; incomplete angle; ill-posed inverse problem

DOI: 10.3969/j.issn.1672-9374.2024.06.001

备注

实现对非定常流动的三维测量一直是航空航天、能源环境、燃烧诊断等领域的研究重点。背景纹影法是近20年来兴起的一种新型流场测量技术,仅需相机和背景就能动态测量非定常流场,具有大视野、高频动态、低成本、好测量的优势。受空间、成本限制可用的测量光路并不多,背景纹影只能在有限角度内稀疏采集。对不完全角度背景纹影层析测量技术的应用现状和发展进行综述:首先介绍背景纹影成像及层析测量的基本原理和测量系统,分析当前背景层析测量中面临的不完全角度问题; 其次回顾近年来所发展的不完全角度背景纹影层析重建算法; 最后对背景纹影层析测量技术的发展提出展望,认为利用压缩感知和其他非端到端式深度学习等理论,结合丰富的先验信息和其他层析技术,可以为不完全角度背景纹影层析重建问题的解决提供新思路。
The three-dimensional measurement of unsteady flow has always been a research focus in fields such as aerospace, energy and environment, and combustion diagnostics. The background-oriented schlieren(BOS)technique has emerged as a novel method for measuring flow dynamics in the past two decades. It allows for the dynamic measurement of unsteady flow using just a camera and a background, offering advantages such as a wide field of view, high-frequency dynamics, cost-effectiveness, and ease of use. However, due to spatial and cost constraints, the available optical paths for BOS are limited, resulting in the sparse collection of background patterns within a restricted angle. This article provides a comprehensive overview of the current status and development of incomplete-angle BOS tomography. It begins by introducing the fundamental principles of BOS imaging and tomographic reconstruction, along with the measurement systems, while analyzing the current incomplete-angle issue encountered in BOS tomographic measurements. Furthermore, it reviews the algorithms for incomplete-angle BOS tomography developed in recent years and concludes by outlining prospects for the future advancement of BOS tomography. It is suggested that leveraging theories such as compressed sensing and other non-end-to-end deep learning approaches, along with rich prior information and other tomographic techniques, could offer new avenues for addressing the reconstruction challenges in the context of incomplete-angle BOS tomography.
·