一种基于双目距离选通的水下层析三维成像方法与系统

沈磊; 裘溯; 金伟其; 钟鑫; 朱世豪; 于相之; 彭光越

doi:10.3788/gzxb20265507.0711001

Chinese

您当前的位置：

首页 >

文章列表页 >

一种基于双目距离选通的水下层析三维成像方法与系统

成像系统 | 更新时间：2026-06-12

- 一种基于双目距离选通的水下层析三维成像方法与系统
- An Underwater Tomographic 3D Imaging Method and System Based on Binocular Range Gating
- 光子学报 2026年页码：1-16
- 作者机构：
  
  1.北京理工大学光电学院光电成像技术与系统教育部重点实验室，北京 100081
  2.中国科学院半导体研究所光电系统实验室，北京 100083
- 作者简介：
  
  沈磊（2001—），男，学生，硕士生，主要研究方向为微光与超宽波段成像。Email: 840488106@qq.com
  裘溯（1973—），男，讲师，博士，主要研究方向为光电成像技术与系统。Email: edmondqiu@bit.edu.cn
- 基金信息：
  
  中关村国家自主创新示范区优化创新创业生态环境支持基金(20230481197);珠海市科技创新专项基金(2220004003077)
- DOI：10.3788/gzxb20265507.0711001
  中图分类号： TN29
- CSTR：32255.14.gzxb20265507.0711001
- 收稿：2026-03-02，
  
  修回：2026-04-03，
  
  录用：2026-04-07，
- 稿件说明：
移动端阅览
沈磊，裘溯，金伟其，等. 一种基于双目距离选通的水下层析三维成像方法与系统［J］.光子学报，2026，55（7）：0711001

SHEN Lei， QIU Su， JIN Weiqi， et al. An Underwater Tomographic 3D Imaging Method and System Based on Binocular Range Gating［J］. Acta Photonica Sinica， 2026， 55（7）：0711001
沈磊，裘溯，金伟其，等. 一种基于双目距离选通的水下层析三维成像方法与系统［J］.光子学报，2026，55（7）：0711001 DOI： 10.3788/gzxb20265507.0711001. CSTR： 32255.14.gzxb20265507.0711001.

SHEN Lei， QIU Su， JIN Weiqi， et al. An Underwater Tomographic 3D Imaging Method and System Based on Binocular Range Gating［J］. Acta Photonica Sinica， 2026， 55（7）：0711001 DOI： 10.3788/gzxb20265507.0711001. CSTR： 32255.14.gzxb20265507.0711001.

摘要

针对现有水下光学成像系统探测距离有限、三维成像精度及其实时性较差的局限，提出了一种基于双目选通成像的水下层析三维成像方法，采用高性能三代双选通像增强型互补金属氧化物半导体+蓝绿脉冲激光器，将激光距离选通成像和双目立体视觉结合，研制了一种基于双目距离选通的水下层析三维成像系统。建立了水下相机成像模型及双目参数标定方法，采用基于颜色恒常性的图像增强算法对左、右图像进行增强，通过基于神经网络的快速立体匹配算法得出稠密视差图，结合视差图和双目参数计算得到深度图，获取图像目标的深度信息，实现水下场景的三维重建。实验表明：系统水下探测距离可达6个水体衰减长度以上，在3个衰减长度下的距离分辨率不低于50 mm，帧频不低于25帧；在水下4.9-5.9 m的探测距离下深度估计误差为10-20 mm，深度差测量误差为2-2.5 mm；在10.19 m，2.83 m，1.95 m三种水体衰减长度下约6.4 m成像距离处的深度估计误差为25-120 mm，深度差测量误差为2.5-6.0 mm。为水下光学成像系统实现远距离三维实时成像提供了一种可行的技术实现方式。

Abstract

With the proposal of China’s maritime strategy， the demand for underwater observation technologies and equipment has become increasingly urgent. Existing underwater optical imaging methods often focus on improving a single aspect， such as imaging range， clarity， 3D reconstruction accuracy， or real-time performance， yet face limitations in achieving a balanced integration of these factors. The three-dimensional detection range remains insufficient， and there are challenges in balancing real-time operation， accuracy， and system control complexity. To address these issues， this study proposes an underwater 3D imaging method that combines range-gated imaging with binocular stereo vision. The aim is to integrate the advantages of range-gated imaging in long-distance tomographic imaging and the efficiency of binocular stereo vision in 3D reconstruction， enabling high-resolution， real-time 3D imaging in complex underwater environments. In particular， the proposed method seeks to simultaneously improve imaging distance， reconstruction accuracy， and computational efficiency within a unified framework.To conduct underwater three-dimensional imaging， we design and develop a binocular range-gated underwater tomographic 3D imaging system. This system primarily consists of a blue-green pulsed laser emission module， a gated imaging module， a power supply and communication transmission module， an optical-electrical composite cable， a pressure-resistant housing， and an upper computer. Each module is integrated to ensure stable operation under underwater pressure conditions and to support underwater communication and real-time data processing. The system achieves gated imaging through the synchronized control of pulsed laser illumination and gated camera exposure， thereby suppressing backscatter and enhancing image contrast. To address the issue of refractive effects at water interfaces during underwater imaging， an underwater camera model that considers the refraction effects at the water-glass-air interface is established. Combined with Zhang Zhengyou's calibration method， underwater camera calibration and correction are implemented. This calibration process ensures accurate estimation of intrinsic and extrinsic parameters， which is essential for reliable depth recovery. For the preprocessed and corrected left and right original underwater images， a Retinex-based adaptive histogram equalization algorithm is applied to enhance contrast. This enhancement step effectively improves the visibility of low-contrast regions and increases the robustness of subsequent stereo matching. A fast neural network-based stereo matching algorithm （BGNet） is then employed to generate dense disparity maps. Finally， using calibrated parameters （reprojection matrix Q）， the disparity maps are converted into depth maps to obtain depth information of the target objects， thereby achieving three-dimensional reconstruction of the underwater scene. The entire processing pipeline is optimized and accelerated to ensure that image enhancement， disparity estimation， and depth reconstruction can be executed efficiently in real time.We conduct underwater two-dimensional imaging experiments and depth reconstruction experiments in both laboratory water tanks and outdoor pools. In the two-dimensional imaging experiments， combined with image enhancement algorithms， the system achieves high-resolution imaging detection of various targets such as USAF 1951 resolution target， fish models， and coral models in the water tank environment. Compared with the original gated images， the enhanced results exhibit significantly improved contrast and clearer structural details. In highly turbid water （with an attenuation length of 1.95 m）， the imaging distance exceeds 6 attenuation lengths. This demonstrates the effectiveness of range-gated imaging in suppressing backscatter and extending the imaging range under severe scattering conditions. In the depth reconstruction experiments， the system performs real-time three-dimensional imaging of different targets in the water tank environment， achieving a frame rate of 25 fps. Within the distance range of 4.9 to 5.9 meters underwater， the depth estimation error is 10-20 mm， and the depth difference measurement error is 2-2.5 mm. In pools with varying turbidity levels （attenuation lengths of 10.19 m， 2.83 m， and 1.95 m）， the system achieves a range resolution better than 5 cm at 3 attenuation lengths， with depth estimation errors of 25-120 mm and depth difference errors of 2.5-6.0 mm. Although the reconstruction accuracy decreases with increasing turbidity， the system still maintains stable depth estimation capability. The experiments demonstrate that even under low illumination and high scattering conditions， the system can effectively reconstruct three-dimensional scenes while maintaining good texture and edge information preservation.The proposed binocular range-gated underwater tomographic 3D imaging system successfully integrates the advantages of range-gated technology for long-distance underwater tomographic imaging with the efficiency and effectiveness of binocular stereo vision in 3D reconstruction. It demonstrates robust performance across various water quality conditions and distance ranges， providing a new research approach and practical solution for long-distance， high-resolution， real-time 3D underwater imaging. The system shows promising application potential in underwater engineering exploration， marine scientific surveys， and dam safety inspections. Future work may focus on optimizing laser beam emission through specialized coupling lenses to homogenize the laser spot， thereby improving underwater illumination uniformity and reducing the impact of lighting variations on accuracy. Additionally， leveraging underwater binocular datasets for training or integrating other stereo matching algorithms will enhance the stereo network’s matching precision， further improving the system’s 3D imaging accuracy in dynamic and turbid underwater environments. Further improvements in hardware integration and algorithm optimization are expected to enhance system stability and broaden its practical application scope.

关键词

Keywords

references

SHEN Ying ， ZHAO Chuanjiang ， LIU Yu ， et al . Underwater optical imaging： key technologies and applications review ［J］. IEEE Access ， 2021 ， 9 ： 85500 - 85514 .

PANETTA K ， GAO Chen ， AGAIAN S . Human-visual-system inspired underwater image quality measures ［J］. IEEE Journal of Oceanic Engineering ， 2016 ， 41 （ 3 ）： 541 - 551 .

WANG Yuhang ， WANG Xinyu ， ZHANG Jinghui ， et al . Review and prospect of underwater 3D imaging detection ［J］. Laser & Optoelectronics Progress ， 2024 ， 61 （ 22 ）： 29 - 42 .

王宇航，王新宇，张景辉，等 . 水下三维光学成像探测技术综述及展望［J］. 激光与光电子学进展， 2024 ， 61 （ 22 ）： 29 - 42 .

ZHUANG Peixian ， WANG Chenyu ， LIU Fei ， et al . Image field-of-view enhancement： An overview ［J］. CAAI Artificial Intelligence Research ， 2026 .

ARST H . Optical properties and remote sensing of multicomponental water bodies ［M］. Berlin ： Springer Science & Business Media ， 2003 .

WANG Xiangke ， ZHANG Hui ， ZHENG Zhiqiang . Analysis of back-scattering restraining of range-gated imaging technique ［J］. High Power Laser and Particle Beams ， 2007 ， 19 （ 12 ）： 1997 - 2000 .

王祥科，张辉，郑志强 . 距离选通技术对后向散射抑制作用的理论分析［J］. 强激光与粒子束， 2007 ， 19 （ 12 ）： 1997 - 2000 .

LI Yifeng ， ZHOU Shun ， LI Pangyue ， et al . Three-dimensional reconstruction method integrating active and passive polarization based on structured light ［J］. Acta Photonica Sinica ， 2024 ， 53 （ 12 ）： 1210002 .

李宜峰，周顺，李庞跃，等 . 基于结构光的偏振主被动融合三维重建方法［J］. 光子学报， 2024 ， 53 （ 12 ）： 1210002 . DOI: 10.3788/gzxb20245312.1210002 http://dx.doi.org/10.3788/gzxb20245312.1210002

ZHUANG Sufeng ， JI Yong ， TU Dawei ， et al . Underwater RGB-D camera based on binocular stereo vision ［J］. Acta Photonica Sinica ， 2022 ， 51 （ 4 ）： 0404003 .

庄苏锋，吉勇，屠大维，等 . 基于双目立体视觉的水下RGB-D相机［J］. 光子学报， 2022 ， 51 （ 4 ）： 0404003 .

ZHUANG Peixian ， ZHANG Xinheng ， TONG Junjie ， et al . Underwater light field imaging： A survey ［J］. CAAI Artificial Intelligence Research ， 2025 ， 4 ： 9150051 .

ZHANG Jinghui ， WANG Yuhang ， ZHANG Tao ， et al . A study on refraction error compensation method for underwater spinning laser scanning three-dimensional imaging ［J］. Sensors ， 2024 ， 24 （ 2 ）： 343 .

LV Wenlei ， ZHANG Xu ， LIU Ke . Design and Experimental Study on a Range-Gated Underwater Laser Imaging System ［J］. Journal of Ordnance Equipment Engineering ， 2019 ， 40 （ 8 ）： 198 - 202 .

吕文磊，张旭，刘可 . 一种距离选通水下激光成像系统设计与实验研究［J］. 兵器装备工程学报， 2019 ， 40 （ 8 ）： 198 - 202 . DOI: 10.11809/bqzbgcxb2019.08.039 http://dx.doi.org/10.11809/bqzbgcxb2019.08.039

MACCARONE A ， DRUMMOND K ， MCCARTHY A ， et al . Submerged single-photon LiDAR imaging sensor used for real-time 3D scene reconstruction in scattering underwater environments ［J］. Optics Express ， 2023 ， 31 （ 10 ）： 16690 - 16708 .

LI Weiwei ， ZHANG Tongyi ， KANG Yan ， et al . Color non-line-of-sight imaging based on time-correlated single photon counting ［J］. Acta Photonica Sinica ， 2024 ， 53 （ 11 ）： 1111001 .

李薇薇，张同意，康岩，等 . 基于时间相关单光子计数的彩色非视域成像［J］. 光子学报， 2024 ， 53 （ 11 ）： 1111001 .

YIN Xiaojun ， CHENG Hu ， YANG Kecheng ， et al . Bayesian reconstruction method for underwater 3D range-gated imaging enhancement ［J］. Applied Optics ， 2020 ， 59 （ 2 ）： 370 - 379 .

JIN Chenfei ， SUN Xiudong ， ZHAO Yuan ， et al . Gain-modulated three-dimensional active imaging with depth-independent depth accuracy ［J］. Optics Letters ， 2009 ， 34 （ 22 ）： 3550 - 3552 .

Walia A ， Walz S ， Bijelic M ， et al . Gated2gated： self-supervised depth estimation from gated images ［C］. IEEE ， 2022 ： 2801 - 2811 .

JIN Dongdong ， PEI Chonglei ， JI Chunheng ， et al . Research on portable underwater laser range gated imaging system ［C］. SPIE ， 2023 ， 12617 ： 126176T .

WU Guanlin ， YU Xiangzhi ， ZHU Shihao ， alet ， Preprocessing method for underwater time-gated SPAD-array-based images utilizing Retinex and BM 3 D ［J］. Applied Optics ， 2025， 64 （ 14 ）， 3842 - 3855 .

LV Sheng ， ZHANG Wenxue ， ZHANG Minghua ， et al . Underwater binocular 3D imaging method based on polarization difference ghost imaging ［J］. Optics and Lasers in Engineering ， 2025 ， 186 ： 108856 .

YamingOU ， FAN Junfeng ， ZHOU Chao ， et al . Binocular structured light 3-D reconstruction system for low-light underwater environments： Design， modeling， and laser-based calibration ［J］. IEEE Transactions on Instrumentation and Measurement ， 2023 ， 72 ： 1 - 14 .

WANG Xinwei ， SUN Liang ， ZHANG Yue ， et al . Advances of laser range-gated three-dimensional imaging （invited）［J］. Infrared and Laser Engineering ， 2024 ， 53 （ 4 ）： 31 - 49 .

王新伟，孙亮，张岳，等 . 激光距离选通三维成像技术研究进展（特邀）［J］. 红外与激光工程， 2024 ， 53 （ 4 ）： 31 - 49 .

STEFANO M . Stereo vision： algorithms and applications ［D］. Bologna ： University of Bologna ， 2013 .

TIPPETTS B ， LEE D J ， LILLYWHITE K ， et al . Review of stereo vision algorithms and their suitability for resource limited systems ［J］. Journal of Real-Time Image Processing ， 2016 ， 11 （ 1 ）： 5 - 25 .

RUSSAKOVSKY O ， DENG Jia ， SU Hao ， et al . Image Net Large Scale Visual Recognition Challenge ［J］. International Journal of Computer Vision ， 2015 ， 115 （ 3 ）： 211 - 252 .

JAFFE J S . Computer modeling and the design of optimal underwater imaging systems ［J］. IEEE Journal of Oceanic Engineering ， 1990 ， 15 （ 2 ）： 101 - 111 .

ZHANG Zhengyou . A flexible new technique for camera calibration ［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence ， 2000 ， 22 （ 11 ）： 1330 - 1334 .

FUSIELLO A ， TRUCCO E ， VERRI A . A compact algorithm for rectification of stereo pairs ［J］. Machine Vision and Applications ， 2000 ， 12 （ 1 ）： 16 - 22 .

ULLAH H ， ZHOU Shun ， DU Kaiyu ， et al . Intrinsic dark-field Fourier ptychographic diffraction tomography under non-matched illumination ［J］. iOptics ， 2025 ， 1 （ 1 ）： 100006 .

LI Dayi ， ZHOU Jingchun ， WANG Shiyin ， et al . Adaptive weighted multiscale retinex for underwater image enhancement ［J］. Engineering Applications of Artificial Intelligence ， 2023 ， 123 ： 106457 .

DING Kun ， HUANG Youwei ， JIN Weiqi ， et al . Experimental study on the relationship between attenuation coefficient of Blue-green laser transmission and the water turbidity ［J］. Infrared Technology ， 2013 ， 35 （ 8 ）： 467 - 471 .

丁琨，黄有为，金伟其，等 . 水下蓝绿激光传输的衰减系数与水体浊度关系的实验研究［J］. 红外技术， 2013 ， 35 （ 8 ）： 467 - 471 .

XU Binfeng ， XU Yuhua ， YANG Xiaoli ， et al . Bilateral Grid Learning for Stereo Matching Networks ［C］. CVPR ， 2021 ： 12492 - 12501 .

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

融合偏振特性的双目立体视觉增强方法

基于自适应截断模拟曝光和无监督融合的低照度真彩色图像增强算法

基于自注意力机制优化的AtG-DeepLab V3+内窥图像增强算法

基于扩展大气散射模型的低光照图像增强算法

基于全局自适应处理的分块影像增强方法