多线程半全局立体匹配工件内部检测方法

doi:10.3969/j.issn.1004-132X.2025.11.037

摘要/Abstract

摘要：

针对传统检测手段难以深入工件内部进行三维量化分析的问题，提出一种基于多线程半全局立体匹配（M-SGSM）的双目视觉检测方法。首先采用张正友标定法获取双目内窥镜参数，并对图像进行畸变矫正与极线矫正；然后设计多线程区域叠加分割策略优化立体匹配算法，提高视差图计算效率，生成三维彩色点云；最后，提出自标定面积测量法结合欧氏空间距离计算实现三维量化分析。实验结果表明，优化后立体匹配算法计算效率提高约30%，所提检测方法生成的三维点云结构清晰，欧氏距离测量误差小于3%，面积测量误差小于1.5%。该方法为工件内部三维检测提供了高效高精解决方案。

关键词: 双目视觉, 立体匹配, 三维重建, 工件测量

Abstract:

To address the problems that traditional detection methods were difficult to conduct three-dimensional quantitative analysis inside workpieces， a binocular vision detection method was proposed based on multi-threaded semi-global stereo matching（M-SGSM）. Firstly， the Zhang's calibration method was used to obtain the parameters of the binocular endoscope， and image distortion correction and epipolar rectification were performed. Secondly， a multi-threaded regional overlap segmentation strategy was designed to optimize the stereo matching algorithm， improve the efficiency of disparity map calculation， and generate a 3D color point cloud. Finally， a self-calibration area measurement method combined with Euclidean distance calculation was proposed to realize three-dimensional quantitative analysis. Experimental results show that the optimized stereo matching algorithm improves computational efficiency by approximately 30%， the generated 3D point cloud has a clear structure， the Euclidean distance measurement error is less than 3%， and the area measurement error is less than 1.5%. This method provides an efficient and high-precision solution for three-dimensional internal detection of workpieces.

Key words: binocular vision, stereo matching, 3D reconstruction, workpiece measurement

中图分类号:

TP391.41

吴洪臣, 陈晓荣, 李柏杨. 多线程半全局立体匹配工件内部检测方法[J]. 中国机械工程, 2025, 36(11): 2783-2791.

Hongchen WU, Xiaorong CHEN, Baiyang LI. Multi-threaded Semi-global Stereo Matching Method for Internal Inspection of Workpieces[J]. China Mechanical Engineering, 2025, 36(11): 2783-2791.

导出引用管理器 EndNote|Ris|BibTeX

链接本文: https://www.cmemo.org.cn/CN/10.3969/j.issn.1004-132X.2025.11.037

https://www.cmemo.org.cn/CN/Y2025/V36/I11/2783

图/表 19

图1 三维重建与测量流程

Fig.1 3D reconstruction and measurement process

图2 双目视觉系统

Fig.2 Binocular vision system

图3 OV9734双目内窥镜

Fig.3 OV9734 binocular endoscope module

图4 双目标定采集图片

Fig.4 Binocular calibration collection images

图5 标定结果

Fig.5 Calibration results

表1 左右相机内外参数与畸变系数

Tab.1 Left and right camera internal and external parameters and distortion coefficients

参数	左相机	右相机
内参矩阵	$296.07 0 316.90 0 295.64 238.46 001$	$294.98 0 319.64 0 294.61 234.93 001$
旋转矩阵	$- 2.518 - 0.053 0.009$
平移矩阵	$0.999 0.004 0.034 - 0.004 0.999 0.013 - 0.034 - 0.013 0.999$
畸变系数	k₁=0.087	k₁=0.068
	k₂=-0.269	k₂=-0.198
	k₃=0.146	k₃=0.080
	p₁=0.000 093	p₁=0.000 249
	p₂=-0.000 984	p₂=-0.001 224

表1 左右相机内外参数与畸变系数

Tab.1 Left and right camera internal and external parameters and distortion coefficients

参数	左相机	右相机
内参矩阵	$296.07 0 316.90 0 295.64 238.46 001$	$294.98 0 319.64 0 294.61 234.93 001$
旋转矩阵	$- 2.518 - 0.053 0.009$
平移矩阵	$0.999 0.004 0.034 - 0.004 0.999 0.013 - 0.034 - 0.013 0.999$
畸变系数	k₁=0.087	k₁=0.068
	k₂=-0.269	k₂=-0.198
	k₃=0.146	k₃=0.080
	p₁=0.000 093	p₁=0.000 249
	p₂=-0.000 984	p₂=-0.001 224

图6 SGBM算法流程

Fig.6 SGBM algorithm process

表2 实验环境

Table.2 Experimental environment

名称	配置信息
操作系统	Windows 11 64位
编程语言	C++ 11
编译器	Microsoft Visual Studio 2017
CPU	Intel Core i7-13700H
GPU	NVIDA RTX 4060 Laptop （8G）

图7 匹配结果

Fig.7 Matching results

表3 SGBM、BM、AD-Census在MiddleBurry数据集上的对比

Tab.3 Comparison of SGBM， BM， AD-Census on the MiddleBurry dataset

测试对象	SSIM			PSNR/dB			计算时间t/ms
测试对象	SGBM	BM	AD-Census	SGBM	BM	AD-Census	SGBM	BM	AD-Census
Art	0.537	0.324	0.826	11.423	8.697	19.936	40.085	3.083	2844
Books	0.687	0.364	0.891	11.571	9.330	21.118	39.667	2.980	2900
Moebius	0.660	0.415	0.882	13.284	10.552	20.696	38.869	3.986	2788
Laundry	0.647	0.281	0.823	11.247	8.917	21.618	37.542	5.233	2660

图8 分割策略

Fig.8 Segmentation strategy

图9 多线程算法实现流程

Fig.9 Multi-threaded algorithm implementation process

表4 各算法在时间上的对比

Tab.4 Comparison of each algorithm in time

测试对象	时间/ms
测试对象	SGBM	BM	AD-Census	M-SGSM
Art	40.085	3.083	2844	23.436
Books	39.667	2.980	2900	28.084
Moebius	38.869	3.986	2788	22.235
Laundry	37.542	5.233	2660	22.797

图10 结构示意

Fig.10 Structure schematic

图12 金属工件

Fig.12 Metal workpiece

图13 上位机软件

Fig.13 PC Software

图14 长度测量示意图

Fig.14 Length measurement

图15 面积测量

Fig.15 Area measurement

表5 三维测量结果

Tab.5 3D measurement results

测量类型	测量对象	实际值	测量结果					平均相对误差/%
测量类型	测量对象	实际值	测量1	测量2	测量3	测量4	测量5	平均相对误差/%
长度测量/mm	等边三角形边长	6.000	5.766	5.958	5.753	5.833	5.996	2.373
	圆直径	8.000	7.917	7.910	7.876	7.923	7.916	1.145
	大三角长边	24.469	23.875	24.063	24.107	24.333	24.134	1.498
	大三角中边	19.256	18.784	18.758	18.763	18.793	19.200	2.058
	大三角短边	18.655	18.588	18.484	18.347	18.352	18.310	1.280
面积测量/mm²	椭圆形	89.512	90.276	90.022	91.309	91.045	90.374	1.221
面积测量/mm²	云朵形	76.751	75.771	75.468	76.235	78.605	77.032	1.281

参考文献 24

[1]	何宝凤，丁思源，魏翠娥，等. 三维表面粗糙度测量方法综述［J］. 光学精密工程， 2019， 27（1）： 78-93.
	HE Baofeng， DING Siyuan， WEI Cuie， et al. Review of Measurement Methods for Areal Surface Roughness［J］. Optics and Precision Engineering， 2019， 27（1）： 78-93.
[2]	那一鸣，胡超，邱业余，等. 基于机器视觉的汽车车门三维定位引导［J］. 中国机械工程， 2024， 35（9）： 1677-1687.
	NA Yiming， HU Chao， QIU Yeyu， et al. Three-dimensional Positioning Guidance of Automobile Doors Based on Machine Vision［J］. China Mechanical Engineering， 2024， 35（9）： 1677-1687.
[3]	易怀安，赵欣佳，唐乐，等.基于彩色图像奇异值熵指标的磨削表面粗糙度视觉测量方法［J］.中国机械工程，2021，32（13）：1577-1583.
	YI Huaian， ZHAO Xinjia， TANG Le， et al. Vision Mearsurement Method for Ground Surface Roughness Based on Color Image Singular Value Entropy Index ［J］. China Mechanical Engineering， 2021， 32（13）： 1577-1583.
[4]	曹旭明，冯浩，曹利钢，等. 基于机器视觉螺栓类零件参数在线质检系统［J］. 控制工程， 2021， 28（6）： 1187-1194.
	CAO Xuming， FENG Hao， CAO Ligang， et al. Online Quality Inspection System for Bolt Parts Parameters Based on Machine Vision ［J］. Control Engineering of China， 2021， 28（6）： 1187-1194.
[5]	马大智，于斌超，张彦泽，等. 基于双目视觉的大型高反光构件测量系统［J］. 应用光学， 2021， 42（4）： 577-585.
	MA Dazhi， YU Binchao， ZHANG Yanze， et al. Measurement System for Large-scale High Reflective Component Based on Binocular Vision［J］. Journal of Applied Optics， 2021， 42（4）： 577-585.
[6]	HughesBaker.Remote Visual Inspection［EB/OL］.Houston： Baker Hughes，2024［2025-06-11］..
[7]	SCHÖLLY FIBEROPTIC GMBH.FLEXIVISO@100 INSPECTIONS［EB/OL］.Freiburg：SCHÖLLY， 2022［2025-06-11］..
[8]	Olympus.IPLEX NX Olympus's Most Advanced Industrial Videoscope［EB/OL］.Tokyo：Olympus，2016［2025-06-11］..
[9]	周富强，王晔昕，柴兴华，等. 镜像双目视觉精密测量技术综述［J］. 光学学报， 2018， 38（8）： 59-67.
	ZHOU Fuqiang， WANG Yexin， CHAI Xinghua， et al. Review on Precise Measurement Technology Based on Mirror Binocular Vision ［J］. Acta Optica Sinica， 2018， 38（8）： 59-67.
[10]	FIRMANSYAH I， YAMAGUCHI Y， TMARU-YAMA， et al. FPGA-Based Stereo Matching for Crop Height Measurement Using Monocular Camera ［J］. Microprocess and Microsystems， 2024， 108： 105063.
[11]	SHAHBAZI M， SOHN G， THEO J. High-density Stereo Image Matching Using Intrinsic Curves ［J］. ISPRS Journal of Photogrammetry and Remote Sensing， 2018， 146： 373-388.
[12]	张泽琳，曹星，王蕾，等. 基于改进SGM的废旧机械零件彩色三维重建方法［J］. 激光与光电子学进展， 2024， 61（12）： 114-124.
	ZHANG Zelin， CAO Xing， WANG Lei， et al. Color 3D Reconstruction Method of Used Machinery Parts Based on Improved SGM ［J］. Laser & Optoelectronics Progress， 2024， 61（12）： 114-124.
[13]	WANG Yina， GU Mengjiao， ZHU Yufeng， et al. Improvement of AD-Census Algorithm Based on Stereo Vision ［J］. Sensors， 2022， 22（18）： 6933.
[14]	GUO Junguang， GU Feifei， YE Yuping， et al. An Accurate Speckle 3D Reconstruction System Based on Binocular Endoscope［C］∥2021 IEEE International Conference on Real-time Computing and Robotics （RCAR）. Xining， 2021： 703-708.
[15]	HAMID M S， MANAP N A， HAMZAH R A， et al. Stereo Matching Algorithm Based on Deep Learning： a Survey ［J］. Journal of King Saud University： Computer and Information Sciences， 2022， 5（34）： 1663-1673.
[16]	YU Hao， ZHOU Changjiang， ZHANG Wei . et al. A Three-dimensional Measurement Method for Binocular Endoscopes Based on Deep Learning［J］. Frontiers of Information Technology & Electronic Engineering， 2022， 23（4）：653-660.
[17]	TAHMASEBI M， HUQ S， MEEHAN K， et al. DCVSMNet： Double Cost Volume Stereo Matching Network ［J］. Neurocomputing， 2024，618： 129002.
[18]	YAN Long， ZHAO Xingfang， DU Huiqiu. Research on 3D Measuring Based Binocular Vision［C］∥2014 IEEE International Conference on Control Science and Systems Engineering. Yantai， 2014： 18-22.
[19]	HUANG Wei， MIAO Huisi， JIAO Shuming， et al. A Planar Constraint Optimization Method to Improve Camera Calibration for Imperfect Planar Targets［J］. Optics and Lasers in Engineering， 2024，180： 108273 .
[20]	ZHANG Xiaowen， Tiegang LYU， DAN Wang， et al.High-precision Binocular Camera Calibration Method Based on a 3D Calibration Object［J］. Applied Optics， 2024， 63（10）： 2667-2682.
[21]	ZHANG Zhengyou. Flexible Camera Calibration by Viewing a Plane from Unknown Orientations［C］ ∥Proceedings of the Seventh IEEE International Conference on Computer Vision. Kerkyra， 1999： 666-673.
[22]	HIRSCHMULLER H. Stereo Processing by Semiglobal Matching and Mutual Information［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence， 2008， 30（2）： 328-341.
[23]	MOSTEGEL C， RUMPLER M， FRAUN-DORFER F， et al. Using Self-contradiction to Learn Confidence Measures in Stereo Vision［C］∥Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas， 2016： 4067-4076.
[24]	MIANGOLEH S M H， DILLE S， MAI L， et al. Boosting Monocular Depth Estimation Models to High-resolution via Content-adaptive Multi-resolution Merging［C］∥ Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Nashville， 2021： 9680-9689.