基于图像特征的双目测距系统的研究

doi:10.16182/j.issn1004731x.joss.20-0793

系统仿真学报 ›› 2022, Vol. 34 ›› Issue (3): 624-632.doi: 10.16182/j.issn1004731x.joss.20-0793

• 仿真支撑平台/系统技术 • 上一篇下一篇

基于图像特征的双目测距系统的研究

杨敬辉¹(), 刘德康¹, 杜万和¹(), 邢立宁²

^1.上海第二工业大学工学部，上海 201209
^2.国防科技大学系统工程学院，湖南长沙 410073

收稿日期:2020-10-16 修回日期:2021-01-11 出版日期:2022-03-18 发布日期:2022-03-22
通讯作者: 杜万和 E-mail:jhyang@sspu.edu.cn;whdu@sspu.edu.cn
作者简介:杨敬辉(1968-)，女，博士，教授，研究方向为ERP、智能制造等。E-mail：jhyang@sspu.edu.cn
基金资助:
上海市一流研究生项目(A30DB201011)

Research on Binocular Ranging System Based on Image Features

Jinghui Yang¹(), Dekang Liu¹, Wanhe Du¹(), Lining Xing²

^1.Engineering Department of Shanghai Polytechnic University, Shanghai 201209, China
^2.School of Systems Engineering, NationalUniversity of Defense Technology, Changsha 410073, China

Received:2020-10-16 Revised:2021-01-11 Online:2022-03-18 Published:2022-03-22
Contact: Wanhe Du E-mail:jhyang@sspu.edu.cn;whdu@sspu.edu.cn

摘要/Abstract

摘要：

针对双目视觉测距中测量误差大、图像信息单一、实时性差等问题，提出一种基于ORB（oriented fast and rotated brief）特征的双目测距方法。对视频帧进行中值滤波处理，提取图像ORB特征，通过实验选出匹配效果最好的汉明距离。对筛选后的匹配点进行RANSAC（random sample consensus）模型估计，去除误匹配，分析视差和真实距离的模型关系，构建最优的测距模型并在实验平台上进行验证。结果表明：所提方法比其他双目测距方法具有测距精确、运行速度快、鲁棒性强的优势，能够实时显示图中特征的距离信息。

关键词: ORB（oriented fast and rotated brief）, 特征匹配, RANSAC（random sample consensus）, 模型优化, 双目测距

Abstract:

Aiming at the problems of large measurement error, single image information, and poor real-time performance in binocular vision ranging, a binocular ranging method based on ORB (oriented fast and rotated brief) features is proposed. Median filtering is performed on the video frame, the ORB feature of the image is extracted, and the Hamming distance with the best matching effect is selected through experiments. The RANSAC (random sample consensus) model estimation is performed on the selected matching points, the mismatches are removed, the model relationship between parallax and true distance is analyzed, the optimal ranging model is constructed and verified on the experimental platform. The results show that the proposed method has the advantages of accurate ranging, fast running speed and strong robustness compared with other binocular ranging methods, and can display the distance information of the features in the image in real time.

Key words: oriented fast and rotated brief（ORB）, feature matching, random sample consensus（RANSAC）, model optimization, binocular ranging

中图分类号:

TP391.41

杨敬辉, 刘德康, 杜万和, 邢立宁. 基于图像特征的双目测距系统的研究[J]. 系统仿真学报, 2022, 34(3): 624-632.

Jinghui Yang, Dekang Liu, Wanhe Du, Lining Xing. Research on Binocular Ranging System Based on Image Features[J]. Journal of System Simulation, 2022, 34(3): 624-632.

图/表 17

图1

图2

图3

表1

左右相机内参

相机	$f$ (pixel)	$k 1$	$k 2$	$k 3$	$P 1$	$P 2$
左	832.320	-0.083 1	0.278 6	-0.258 3	-0.002 8	0.002 2
右	832.320	-0.054 7	0.128 9	-0.086 3	0	0.001 4

表1

图4

图5

图6

表2

图7

表3

图8

图9

表4

3种拟合方式的R2，SSE，RMSE对比

拟合方式	$R 2$	SSE	RMSE
高斯拟合	0.999 3	4 166	14.08
多项式拟合	0.999 6	1 975	9.071
幂函数拟合	0.999 9	617	4.694

表4

图10

图11

图12

图13

参考文献 15

1	赵晨园, 李文新, 张庆熙. 双目视觉的立体匹配算法研究进展[J]. 计算机科学与探索, 2020, 14(7): 1104-1113.
	Zhao Chenyuan, Li Wenxin, Zhang Qingxi. Research and Development of Binocular Stereo Matching Algorithm[J]. Journal of Frontiers of Computer Science & Technology, 2020, 14(7): 1104-1113.
2	陈立功, 李柯达, 张泽宇, 等. 基于计算机视觉的目标测距算法[J]. 弹箭与制导学报, 2020, 40(2): 93-96.
	Chen Ligong, Li Keda, Zhang Zeyu, et al. Research on Target Ra-nging Algorithms Based on Computer Vision[J].Journal of Projectiles, Rockets, Missiles and Guidance, 2020, 40(2): 93-96.
3	谢志宏, 王楠洋, 宗艳桃, 等. 基于图像分割的双目无源测距[J]. 计算机仿真, 2019, 36(2): 250-254.
	Xie Zhihong, Wang Nanyang, Zong Yantao, et al. Binocular Passive Ranging Based on Image Segmentation[J]. Computer Integrated Manufacturing Systems, 2019, 36(2): 250-254.
4	杨晨曦, 华云松. 基于双目立体视觉的目标物测距研究[J]. 软件, 2020, 41(1): 128-132.
	Yang Chenxi, Hua Yunsong. Research on Target Ranging Based on Binocular Stereo Vision[J]. Computer Engineering & Software, 2020, 41(1): 128-132.
5	张志强, 施文华. 改进的尺度不变特征变换算法并行加速双目测距系统及其实现[J]. 激光与光电子学进展,2019, 56(14): 201-210.
	Zhang Zhiqiang, Shi Wenhua. Research and Implementation of Binocular Distance Measurement System Based on Improved Scale-Invariant Feature Transform Algorithm with Parallel Acceleration[J]. Laser & Optoelectro-nics Progress, 2019, 56(14): 201-210.
6	顾金凤, 刘祥勇, 纪亚强, 等. 基于视差几何模型的立体视觉测距方法[J]. 江苏科技大学学报(自然科学版),2018, 32(6): 809-815.
	Gu Jinfeng, Liu Xiangyong, Ji Yaqiang, et al. Stereo Vision Ranging Method Based on Parallax Geometric Model[J]. Journal of Jiangsu University of Science and Technology(Natural Science Edition),2018, 32(6): 809-815.
7	蒋晓东, 于纪言, 朱立坤, 等. 基于硬件SURF算法的自校准双目测距系统[J]. 光学学报, 2018, 38(10): 408-410.
	Jiang Xiaodong, Yu Jiyan, Zhu Likun, et al. Self-Calibrated Binocular Ranging System Based on Hardware SURF Algorithm[J]. Acta Optica Sinica,2018, 38(10): 408-410.
8	易靖宇. 基于双目视频采集的深度图像提取与分析[D].哈尔滨: 哈尔滨工业大学, 2019.
	Yi Jingyu. Depth Image Extraction And Analysis Based on Binocular Video Capture[D]. Harbin: Harbin Institute of Technology, 2019.
9	邵暖, 李惠光, 刘乐. 基于特征匹配算法的双目视觉测距[J]. 燕山大学学报, 2012, 36(1): 57-61.
	Shao Nuan, Li Huiguang, Liu Le. Binocular Stereo Distance-Measurement Method Based On Feature Matching Algorithm[J]. Journal of Yanshan University, 2012, 36(1): 57-61.
10	Mur-Artal R, Tardós J D. ORB-SLAM2: An Open-Source SLAM System for Monocular,Stereo,and RGBD Cameras[J].IEEE Transactions on Robotics (S1552-3098), 2017, 33(5): 1255-1262.
11	刘诗婷, 金伟其, 李力, 等. 基于双目变焦超分辨率成像的测距方法[J]. 光学学报, 2020, 40(14): 98-106.
	Liu Shiting, Jin Weiqi, Li Li, et al. Ranging Method Based on Binocular Zoom Super-Resolution Imaging[J]. Acta Optica Sinica, 2020, 40(14): 98-106.
12	孙胜男, 吕玉祥, 王帅, 等. 基于双目测距的输电线交叉跨越测量[J]. 高电压技术, 2018, 44(11): 3657-3663.
	Sun Shengnan, Yuxiang Lü, Wang Shuai,et al. Measurement of Space Crossing Transmission Line Based on Binocular Ranging[J]. High Voltage Engineering, 2018, 44(11): 3657-3663.
13	王磊, 李会容, 周科, 等. 嵌入式ARM的果蔬采摘双目视觉系统与算法设计[J]. 光电子·激光, 2020, 31(1):71-80.
	Wang Lei, Li Huirong, Zhou Ke, et al. Design of Binocular Vision System for Fruit and Vegetable Picking Based on Embeddedarm[J]. Journal of Optoelectronics.Laser, 2020, 31(1): 71-80.
14	史珂路, 田军委, 雷志强, 等. 基于融合动态模板匹配的双目测距算法[J]. 计算机系统应用, 2018, 27(11): 218-223.
	Shi Kelu, Tian Junwei, Lei Zhiqiang, et al. Binocular Ranging Algorithm Based on Fusion Dynamic Template Matching[J]. Computer Systems & Applications, 2018, 27(11): 218-223.
15	安果维, 王耀南, 周显恩, 等. 基于显著性检测的双目测距系统[J]. 智能系统学报, 2018, 13(6): 913-920.
	An Guowei, Wang Yaonan, Zhou xianen,et al. Binocular Distance Measurement System Based on Saliency Detection[J]. CAAI Transactions on Intelligent Systems, 2018, 13 (6): 913-920.

汉明距离	总点数	筛选后点数	RANSAC剔除
20	1 000	141	131
25	1 000	221	195
30	1 000	331	294
35	1 000	422	364
40	1 000	501	413
45	1 000	569	414

z/mm	d(pixel)	z/mm	d(pixel)	z/mm	d(pixel)
300	355.368	800	128.099	1 300	77.895
350	300.087	850	119.862	1 350	74.850
400	259.303	900	113.026	1 400	71.942
450	230.157	950	107.113	1 450	69.765
500	206.663	1 000	101.946	1 500	67.679
550	186.678	1 050	97.022	1 550	64.846
600	171.513	1 100	92.237	1 600	63.482
650	157.999	1 150	88.013	1 650	61.070
700	146.544	1 200	84.834	1 700	59.866
750	135.873	1 250	80.920	1 750	57.554

[1]	宁小娟, 李洁茹, 高凡, 王映辉. 基于最佳几何约束和RANSAC的特征匹配算法[J]. 系统仿真学报, 2022, 34(4): 727-734.
[2]	王红微, 杨鹏. 基于深度强化学习的机场货运业务优化研究[J]. 系统仿真学报, 2022, 34(3): 651-660.
[3]	蔡鹏, 沈朝萍, 李红燕. 基于标定区域特征点组合匹配的位姿估计方法[J]. 系统仿真学报, 2021, 33(7): 1638-1646.
[4]	张友鹏, 王淳, 刘艳丽. 移动视点下在线视频的动态阴影检测与跟踪[J]. 系统仿真学报, 2019, 31(7): 1439-1447.
[5]	孙晶晶, 于佳骏, 李贝, 谢志峰, 丁友东. 基于人像字典集的卡通自动生成方法[J]. 系统仿真学报, 2015, 27(4): 682-688.

基于图像特征的双目测距系统的研究

Research on Binocular Ranging System Based on Image Features

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 17

参考文献 15

相关文章 5

编辑推荐

Metrics

本文评价