一种基于图像处理技术的镜框尺寸测量系统

doi:10.16180/j.cnki.issn1007-7820.2022.11.007

摘要/Abstract

摘要：

针对在特殊情况下传统镜框尺寸测量方法存在不适用性的问题,文中提出了一种基于图像的非接触测量方法,利用图像处理技术对镜框轮廓进行提取测量。采用ERT定位人脸并根据特征点定位人脸镜框区域,在预处理和轮廓检测过程中使用改进的Canny算法得到更完整的镜框轮廓。镜框识别模式采用最小外接矩形法进行校正,提高了对镜框的提取精度,也便于后续对镜框轮廓进行尺寸测量。以瞳距为参照物得到镜框轮廓的实际尺寸信息。PC端的算法实现结果表明,文中所提的测量算法复杂度小、精确度高、测量速度快,可以在移动设备上对镜框轮廓尺寸进行快速准确的测量。

关键词: 图像处理, 镜框尺寸, 轮廓检测, ERT, 自适应, PD, 非接触, 预处理

Abstract:

In response to the inapplicability of traditional frame size measurement methods under special circumstances, this study proposes an image-based non-contact measurement method which uses image processing technology to extract and measure the outline of the frame. ERT is used to locate the face and locate the face frame area according to the feature points. In the process of preprocessing and contour detection, an improved Canny algorithm is used to obtain a more complete frame contour. The frame recognition mode is calibrated by the minimum bounding rectangle method, which improves the extraction accuracy of the frame, and facilitates the subsequent size measurement of the frame outline. The actual size information of the outline of the lens frame is obtained with the interpupillary distance as the reference object. The results of the algorithm implementation on the PC side show that the proposed measurement algorithm has low complexity, high accuracy, and fast measurement speed, and can quickly and accurately measure the outline size of the frame on a mobile device.

Key words: image processing, frame size, contour detection, ERT, adaptive, PD, non contact, pretreatment

中图分类号:

TP391

张志浩,付东翔,闫蕊. 一种基于图像处理技术的镜框尺寸测量系统[J]. 电子科技, 2022, 35(11): 42-47.

ZHANG Zhihao,FU Dongxiang,YAN Rui. A Measuring System for Frame Size Based on Image Processing Technology[J]. Electronic Science and Technology, 2022, 35(11): 42-47.

图/表 12

图1

图2

图3

图4

图5

图6

图7

图8

图9

表1

表2

表3

参考文献 18

[1]	徐建华. 图像处理与分析[M]. 北京: 科学出版社, 1992.
	Xu Jianhua. Graphic processing and analysis[M]. Beijing: Science Press, 1992.
[2]	李鹃. 基于特征点定位的虚拟试戴的研究[D]. 上海: 上海交通大学, 2011.
	Li Juan. Research on virtual try-on based on feature points extraction[D]. Shanghai: Shanghai Jiao Tong University, 2011.
[3]	张洪, 周益华, 吴静静. 基于改进Canny和异常轮廓点搜索的密封圈毛刺检测系统设计[J]. 组合机床与自动化加工技术, 2020(10):126-130.
	Zhang Hong, Zhou Yihua, Wu Jingjing. Design of burr detection system of sealing ring based on improved canny and abnormal contour point search[J]. Modular Machine Tool and Automatic Manufacturing Technology, 2020(10):126-130.
[4]	王雅静, 宋丹, 陈晓玲. 基于改进Canny算法的女性服装款式轮廓检测研究[J]. 湖南工程学院学报(自然科学版), 2020, 30(2):62-66.
	Wang Yajing, Song Dan, Chen Xiaoling. Research on edge detection of women's garment style based on improved canny algorithms[J]. Journal of Hunan Institute of Technology(Natural Science Edition), 2020, 30(2):62-66.
[5]	樊飞. 人脸特征点检测算法的改进与疲劳检测算法的研究[D]. 南京: 东南大学, 2019.
	Fan Fei. Improvement of facial landmarks detection algorithm along with research on real-time drowsiness detection algorithm[D]. Nanjing: Southeast University, 2019.
[6]	Pan Z X, Tian S, Guo M Z, et al. Comparison of medical image 3D reconstruction rendering methods for robot-assisted surgery[C]. Hefei: Proceedings of the Second International Conference on Advanced Robotics and Mechatronics, 2017.
[7]	田明锐, 孙宏斐, 胡永彪. 一种新的基于Canny算子的道路标线检测算法[J]. 电子设计工程, 2021, 29(1):16-19.
	Tian Mingrui, Sun Hongfei, Hu Yongbiao. A new road marking detection algorithm based on Canny operator[J]. Electronic Design Engineering, 2021, 29(1):16-19.
[8]	刘芬. 基于线性和非线性算法的高效人脸识别系统[J]. 电子科技, 2019, 32(7):82-86.
	Liu Fen. An efficient face recognition system based on linear and nonlinear algorithms[J]. Electronic Science and Technology, 2019, 32(7):82-86.
[9]	Kobayashi T, Sungiura Y, Saito H, et al. Automatic eyeglasses replacement for a 3D virtual try-on system[C]. Reims: Proceedings of the Tenth Augmented Human International Conference, 2019.
[10]	郭杭, 付东翔, 项华中. 基于平行光管的渐进多焦点眼镜片测量系统设计[J]. 电子科技, 2018, 31(4):9-11.
	Guo Hang, Fu Dongxiang, Xiang Huazhong. The measurement system design of progressive addition lens's focal power and astigmatism based on collimator[J]. Electronic Science and Technology, 2018, 31(4):9-11.
[11]	管栋. 基于结构光的眼镜镜框轮廓测量系统的研究[D]. 杭州: 浙江大学, 2017.
	Guan Dong. Research on spectacle frame measurement system based on structured light method[D]. Hangzhou: Zhejiang University, 2017.
[12]	崔凌燕. 基于机器视觉的眼镜镜片轮廓提取系统[D]. 太原: 中北大学, 2018.
	Cui Lingyan. The contour extraction system of glass lens based on machine vision[D]. Taiyuan: North University of China, 2018.
[13]	李健, 刘孔宇, 任宪盛, 等. 基于自适应阈值的Canny算法在MRI边缘检测中的应用[J]. 吉林大学学报(工学版), 2021, 51(2):712-719.
	Li Jian, Liu Kongyu, Ren Xiansheng. Application of canny algorithm based on adaptive threshold in MRI edge detection[J]. Journal of Jilin University(Engineering and Technology Edition), 2021, 51(2):712-719.
[14]	刘成刚, 刘二林, 姜香菊. 改进Canny算子的列车轮对踏面边缘检测算法[J]. 铁道机车车辆, 2021, 41(1):22-25.
	Liu Chenggang, Liu Erlin, Jiang Xiangju. Edge detection algorithm of train wheelset tread based on improved Canny operator[J]. Railway Locomotive & Car, 2021, 41(1):22-25.
[15]	尚小强, 陈从桂, 欧阳光. 基于机器视觉的轮廓度检测算法研究[J]. 机电工程技术, 2019, 48(1):21-23.
	Shang Xiaoqiang, Chen Conggui, Ouyang Guang. Research on contour detection algorithm based on machine vision[J]. Mechanical and Electrical Engineering Technology, 2019, 48(1):21-23.
[16]	许鑫杰, 王秀锋, 鲁文其, 等. 基于边缘检测的零件轮廓识别系统开发[J]. 机电工程, 2019, 36(2):201-205.
	Xu Xinjie, Wang Xiufeng, Lu Wenqi, et al. Development of part contour recognition system based on edge detection[J]. Journal of Mechanical & Electrical Engineering, 2019, 36(2):201-205.
[17]	翟伟良, 姜立标, 姚江波, 等. 边缘轮廓圆的测距及变形检测算法设计[J]. 机械设计与制造工程, 2019, 48(12):85-89.
	Zhai Weiliang, Jiang Libiao, Yao Jiangbo. Algorithm of rangingand deformation detection for the edge contour circle[J]. Mechanical Design and Manufacturing Engineering, 2019, 48(12):85-89.
[18]	漆世钱. 基于轮廓识别和BGR颜色空间的车牌定位[J]. 计算机技术与发展, 2020, 30(12):176-180.
	Qi Shiqian. License plate location based on contour recognition and BGR color space[J]. Computer Technology and Development, 2020, 30(12):176-180.

误差范围 /mm	合格张数	不合格张数	合格率 /%
±0.5	217	83	72.3
±1.0	256	44	85.3
±1.5	278	22	92.7
±2.0	291	9	97.0

误差范围 /mm	合格张数	不合格张数	合格率 /%
±0.5	153	147	51.0
±1.0	188	112	62.7
±1.5	238	62	79.3
±2.0	263	37	87.7

目标	对比	DBL /mm	HBOX /mm	VBOX /mm	SEGHT /mm	ED /mm
1	测量值	26.8	53.0	43.6	11.4	57.7
1	实际值	27.9	55.1	45.7	13.4	57.8
2	测量值	26.5	73.9	64.4	14.2	66.4
2	实际值	26.6	71.7	65.5	15.1	67.7
3	测量值	12.1	41.0	62.6	13.4	62.2
3	实际值	13.3	41.7	63.7	14.5	62.7
4	测量值	20.1	34.2	61.5	9.9	60.0
4	实际值	19.2	34.7	62.1	10.1	59.4