Suzhou Electric Appliance Research Institute
期刊號(hào): CN32-1800/TM| ISSN1007-3175

Article retrieval

文章檢索

首頁(yè) >> 文章檢索 >> 文章瀏覽排名

基于3D點(diǎn)云數(shù)據(jù)的產(chǎn)品缺陷檢測(cè)研究

來(lái)源:電工電氣發(fā)布時(shí)間:2023-02-06 14:06 瀏覽次數(shù):829

基于3D點(diǎn)云數(shù)據(jù)的產(chǎn)品缺陷檢測(cè)研究

李潮林1,陳仲生1,2,左旺1,侯幸林2
(1 湖南工業(yè)大學(xué) 電氣與信息工程學(xué)院,湖南 株洲 412007;
2 常州工學(xué)院 汽車(chē)工程學(xué)院,江蘇 常州 213032)
 
    摘 要:傳統(tǒng) 2D 視覺(jué)檢測(cè)技術(shù)存在效率低下、檢測(cè)精確度較低等不足,3D 視覺(jué)技術(shù)因能顯著提高缺陷檢測(cè)的效率和可靠性得到了高度關(guān)注和廣泛研究。對(duì)已有文獻(xiàn)進(jìn)行了廣泛調(diào)研分析,介紹了 3D 點(diǎn)云數(shù)據(jù)的基本概念、獲取方式及其預(yù)處理方法,重點(diǎn)歸納了傳統(tǒng)點(diǎn)云數(shù)據(jù)缺陷檢測(cè)方法和點(diǎn)云數(shù)據(jù)深度學(xué)習(xí)缺陷檢測(cè)方法,并探討了當(dāng)前研究中存在的問(wèn)題與挑戰(zhàn)。
    關(guān)鍵詞: 3D 視覺(jué);缺陷檢測(cè);點(diǎn)云數(shù)據(jù)
    中圖分類(lèi)號(hào):TP391.41     文獻(xiàn)標(biāo)識(shí)碼:A     文章編號(hào):1007-3175(2023)01-0048-07
 
Research on Product Defect Detection Based on 3D Point Cloud Data
 
LI Chao-lin1, CHEN Zhong-sheng1,2, ZUO Wang1, HOU Xing-lin2
(1 School of Electrical and Information Engineering, Hunan University of Technology, Zhuzhou 412007, China;
2 School of Automotive Engineering, Changzhou Institute of Technology, Changzhou 213032, China)
 
    Abstract: The traditional 2D vision detection technology has disadvantages of low efficiency and detection accuracy, while 3D vision technology can significantly improve its detection efficiency and reliability, so it has been paid high attention and widely analyzed. After making extensive analysis of the existing literature, the paper introduces the basic concept, access and pretreatment method of 3D point cloud data, summarizes the traditional point cloud data defect detection method and point cloud data deep learning defect detection method, and finally discusses the problems and challenges of the current research.
    Key words: 3D vision; defect detection; point cloud data
 
參考文獻(xiàn)
[1] LI Yin, SONG Yuanjia, YANG Zhengwei, et al.Use of line laser scanning thermography for the defect detection and evaluation of composite material[J].Science and Engineering of Composite Materials,2022,29(1):74-83.
[2] 肖克來(lái)提. 表面缺陷檢測(cè)應(yīng)用研究綜述[J] . 電子技術(shù),2020,49(8):189-191.
[3] REN Z, FANG F, YAN N, et al.State of the art in defect detection based on machine vision[J].International Journal of Precision Engineering and Manufacturing-Green Technology,2022,9(2):661-691.
[4] DONG S, WANG P, ABBAS K.A survey on deep learning and its applications[J].Computer Science Review,2021,40:100379.
[5] MING Wuyi, ZHANG Shengfei, LIU Xuewen, et al.Survey of Mura Defect Detection in Liquid Crystal Displays Based on Machine Vision[J].Crystals,2021,11(12):1444.
[6] WANG Yangfan, WANG Chen, LONG Peng, et al.Recent advances in 3D object detection based on RGB-D:A survey[J].Displays,2021,70:102077.
[7] 楊佳琪,張世坤,范世超,等. 多視圖點(diǎn)云配準(zhǔn)算法綜述[J] . 華中科技大學(xué)學(xué)報(bào)(自然科學(xué)版),2022,50(11):16-34.
[8] 韓瑞路. 航空發(fā)動(dòng)機(jī)葉片類(lèi)零件三維重建與缺陷檢測(cè)關(guān)鍵技術(shù)研究[D]. 北京:北方工業(yè)大學(xué),2021.
[9] 劉陽(yáng)陽(yáng). 三維點(diǎn)云數(shù)據(jù)預(yù)處和分割算法的研究[D].西安:西安工程大學(xué),2019.
[10] 楊宜林,李積英,王燕,等. 基于 NDT 和特征點(diǎn)檢測(cè)的點(diǎn)云配準(zhǔn)算法研究[J] . 激光與光電子學(xué)進(jìn)展,2022,59(8):198-204.
[11] 宋成航,李晉儒,劉冠杰. 利用特征點(diǎn)采樣一致性改進(jìn) ICP 算法點(diǎn)云配準(zhǔn)方法[J]. 北京測(cè)繪,2021,35(3):317-322.
[12] 周亞男,喬勛. 基于逆向工程的三維激光掃描點(diǎn)云數(shù)據(jù)濾波方法[J]. 激光雜志,2021,42(9):170-174.
[13] EGUCHI M, KAWAMURA A, TOMIYAMA K, et al.A simplified method of detecting spot surface defects by using quasi-3D data from a conventional road profiler[J].Transportation Research Record,2019,2673(11):377-387.
[14] 朱秀敏,黃磊. 基于三維激光點(diǎn)云的零件表面缺陷檢測(cè)[J]. 儀表技術(shù)與傳感器,2022(7):56-60.
[15] LUO Lufeng, YIN Wei, NING Zhengtong, et al.In-field pose estimation of grape clusters with combined point cloud segmentation and geometric analysis[J].Computers and Electronics in Agriculture,2022,200:107197.
[16] 顏廷鈺. 基于點(diǎn)云的高精度測(cè)量與缺陷檢測(cè)[D] .南京:南京理工大學(xué),2019.
[17] 劉永治. 基于線激光掃描的零件三維表面缺陷檢測(cè)[D]. 西安:西安工程大學(xué),2021.
[18] CHU H H, WANG Z Y.A vision-based system for post-welding quality measurement and defect detection[J].The International Journal of Advanced Manufacturing Technology,2016,86(9):3007-3014.
[19] ZHANG D, ZOU Q, LIN H, et al.Automatic pavement defect detection using 3D laser profiling technology[J] . Automation in Construction,2018,96 :350-365.
[20] 羅宏亮. 基于點(diǎn)云特征的高鐵重軌表面缺陷三維輪廓測(cè)量方法[D]. 沈陽(yáng):東北大學(xué),2018.
[21] HONGSEOK P, MANI T U.Development of an inspection system for defect detection in pressed parts using laser scanned data[J].Procedia Engineering,2014,69:931-936.
[22] XIONG Z, LI Q, MAO Q, et al.A 3D laser profiling system for rail surface defect detection[J].Sensors,2017,17(8):1791.
[23] GUO M, SUN M, PAN D, et al.High-precision detection method for large and complex steel structures based on global registration algorithm and automatic point cloud generation[J].Measurement 2021,172 :108765.
[24] 宋淑雅. 基于改進(jìn)歐式聚類(lèi)的點(diǎn)云分割方法[J] .計(jì)量與測(cè)試技術(shù),2022,49(5):96-100.
[25] 李留昭,皇攀凌,周軍,等. 多區(qū)域分割的三維激光點(diǎn)云障礙物檢測(cè)與應(yīng)用[J] . 激光雜志,2022,43(8):66-70.
[26] HUI T W, PANG G K H.Solder paste inspection using region-based defect detection[J].The International Journal of Advanced Manufacturing Technology,2009,42(7) :725-734.
[27] ZONG Yulong, JIN Liang, WANG Huan, et al.An intelligent and automated 3D surface defect detection system for quantitative 3D estimation and feature classification of material surface defects[J].Optics and Lasers in Engineering,2021,144 :106633.
[28] LIU W, LIU Z, LI Q, et al.High-precision detection method for structure parameters of catenary cantilever devices using 3-D point cloud data[J].IEEE Transactions on Instrumentation and Measurement,2020,70:1-11.
[29] 李炳臻,姜文志,顧佼佼,等. 基于卷積神經(jīng)網(wǎng)絡(luò)的目標(biāo)檢測(cè)算法綜述[J] . 計(jì)算機(jī)與數(shù)字工程,2022,50(5):1010-1017.
[30] ZHANG R, WU Y, ZHANG G, et al.Study on Huizhou architecture of point cloud registration based on optimized ICP algorithm[C]//IOP Conference Series:Earth and Environmental Science,2018.
[31] QI C R, YI L, SU H, et al.Pointnet++ :Deep Hierarchical Feature Learning on Point Sets in a Metric Space[C]// NIPS,2017.
[32] 張建民,陳富健,龍佳樂(lè). 基于圖像處理的點(diǎn)云濾波算法[J] . 激光與光電子學(xué)進(jìn)展,2021,58(6):229-240.
[33] LEE J H, OH H M, KIM M Y.Deep learning based 3D defect detection system using photometric stereo illumination[C]//2019 International Conference on Artificial Intelligence in Information and Communication(ICAIIC),2019.
[34] 陳亮. 基于 3D 視覺(jué)的輪胎成型缺陷檢測(cè)[D]. 青島:青島理工大學(xué),2020.
[35] 王磊. 基于光度立體的金屬板帶表面缺陷三維檢測(cè)方法[D]. 北京:北京科技大學(xué),2019.
[36] 鞠皋林. 基于卷積神經(jīng)網(wǎng)絡(luò)的 PCB 焊錫三維點(diǎn)云數(shù)據(jù)的缺陷檢測(cè)[D]. 上海:華東師范大學(xué),2022.
[37] EDRIS M Z B , JAWAD M S , ZAKARIA Z .Surface defect detection and neural network recognition of automotive body panels[C]//2015 IEEE International Conference on Control System, Computing and Engineering(ICCSCE),2015.
[38] WU K, TAN J, LI J, et al.Few-shot learning approach for 3D defect detection in lithium battery[C]//Journal of Physics :Conference Series,2021.
[39] ZHOU Y , TUZEL O . Voxelnet:End-to-end learning for point cloud based 3d object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2018.
[40] YAN Y, MAO Y, LI B.Second :Sparsely embedded convolutional detection[J].Sensors,2018,18(10):3337.
[41] SHI S, GUO C, JIANG L, et al.Pv-rcnn:Pointvoxel feature set abstraction for 3d object detection[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,2020.
[42] NASROLLAHI M, BOLOURIAN N, HAMMAD A.Concrete surface defect detection using deep neural network based on lidar scanning[C]//Proceedings of the CSCE Annual Conference,2019.
[43] 于浩. 基于激光掃描點(diǎn)云深度學(xué)習(xí)的斜軋穿孔機(jī)頂頭缺陷在線檢測(cè)[D]. 秦皇島:燕山大學(xué),2021.
[44] YAN Z, SHI B, SUN L, et al.Surface defect detection of aluminum alloy welds with 3D depth image and 2D gray image[J] .The International Journal of Advanced Manufacturing Technology,2020,110(3) :741-752.
[45] BESL P J, MCKAY N D.A method for registration of 3-D shapes[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1992,14(2):239-256.