Monocular Dynamic Machine Vision-Based Pearl Shape Detection

doi:10.1007/s12204-019-2103-5

Abstract

Abstract: In terms of the requirement of automatically sorting pearls, the pearl contour feature extraction and shape recognition algorithm are studied in this paper to reckon with the rapid identification of pearls shape online, and a monocular dynamic machine vision-based pearl shape detection device is designed. Through blowing, the pearl is suspended in a funnel shaped container and flipped rapidly in the device. The entire surface image of the pearl to be measured can be promptly grasped by the camera placed right above the funnel. The results of illumination experiments conducted from different angles indicate that the image contour acquired by the medium angle illumination is better extracted. The pearl shape test indicates that the method is incorporated with the inflatable suspension device to classify the pearls into seven types according to the national standard, and additionally the average error rate is confined under 5.38%. The shape characteristic of the pearl can be detected promptly and reliably, and accordingly the high-speed automatic sorting can be satisfied.

Key words: pearl| machine vision| monocular dynamic| suspension| shape characteristics

摘要： In terms of the requirement of automatically sorting pearls, the pearl contour feature extraction and shape recognition algorithm are studied in this paper to reckon with the rapid identification of pearls shape online, and a monocular dynamic machine vision-based pearl shape detection device is designed. Through blowing, the pearl is suspended in a funnel shaped container and flipped rapidly in the device. The entire surface image of the pearl to be measured can be promptly grasped by the camera placed right above the funnel. The results of illumination experiments conducted from different angles indicate that the image contour acquired by the medium angle illumination is better extracted. The pearl shape test indicates that the method is incorporated with the inflatable suspension device to classify the pearls into seven types according to the national standard, and additionally the average error rate is confined under 5.38%. The shape characteristic of the pearl can be detected promptly and reliably, and accordingly the high-speed automatic sorting can be satisfied.

关键词: pearl| machine vision| monocular dynamic| suspension| shape characteristics

CLC Number:

TP 391.4

WANG Yuzong (王毓综), DENG Fei (邓飞), ZHAO Daxu (赵大旭), YE Jiaying (叶佳英), WANG Peixin (王佩欣), SHOU Guozhong (寿国忠). Monocular Dynamic Machine Vision-Based Pearl Shape Detection[J]. Journal of Shanghai Jiao Tong University (Science), 2019, 24(5): 654-662.

References 29

[9]	YANG Z Y, ZHANG W Q, LI W, et al. Informationacquisition method of potted-seedling transplantingfitness using monocular vision [J]. Transactions ofthe Chinese Society of Agricultural Engineering, 2014,30(3): 112-119 (in Chinese).
[10]	ZHOU Z, HUANG Y, LI X Y, et al. Automatic detectingand grading method of potatoes based on machinevision [J]. Transactions of the Chinese Societyof Agricultural Engineering, 2012, 28(7): 178-183 (inChinese).
[1]	ZHENG C H, HUANG L, TANG W, et al. Research onpearl shape and size detection system based on imageprocessing [J]. Journal of China University of Metrology,2014, 25(3): 258-262 (in Chinese).
[11]	LI G, LI B, WANG Y, et al. Pearl shape recognitionbased on computer vision [J]. Transactions of the ChineseSociety for Agricultural Machinery, 2008, 39(7):129-132 (in Chinese).
[2]	HUANG L, ZHENG C H, ZHOU M L, et al. Researchon pearl shape and thread detection system based onimage processing [J]. Science and Technology InnovationHerad, 2014, 11(8): 65-66 (in Chinese).
[3]	LI Y H, LU Z Y, LI T J. New pearl shape symmetrydetection algorithm [J]. Computer Engineering andApplications, 2012, 48(24): 37-40 (in Chinese).
[12]	TANG Y P, XIA S J, FENG Y J, et al. Pearl onlinedetecting and grading device based on monocularmulti-view machine vision [J]. Transactions of the ChineseSociety for Agricultural Machinery, 2014, 45(1):288-292 (in Chinese).
[13]	WANG YW, QU G T, LIU X L, et al. Image subtractiondetection algorithm for surface defect of steel ball[J]. Journal of Computer-Aided Design & ComputerGraphics, 2016, 28(10): 1699-1704 (in Chinese).
[4]	ZHENG H W, CAO Y L, YANG J X. Separationmethod for shape grading of pearls using computer vision[J]. Journal of Engineering Design, 2008, 15(5):365-368 (in Chinese).
[5]	UNAY D, GOSSELIN B, KLEYNEN O, et al. Automaticgrading of bi-colored apples by multispectralmachine vision [J]. Computers and Electronics in Agriculture,2011, 75(1): 204-212.
[14]	SUN H, WANG Z, FU L H, et al. Analysis andimplementation of novel method for steel ball surfacedetection [J]. Mechanical Science and Technologyfor Aerospace Engineering, 2016, 35(1): 118-121 (inChinese).
[6]	RIQUELME M T, BARREIRO P, RUIZ-ALTISENTM, et al. Olive classification according to external damageusing image analysis [J]. Journal of Food Engineering,2008, 87(3): 371-379.
[15]	QUAN Y M, ZHU G Q, JIANG C C, et al. Makingsmall bell welter for automatic detection of surface defect[J]. Modern Manufacturing Engineering, 2010 (9):115-117 (in Chinese).
[16]	Standardization Administration of the P.R.C. Culturedpearl grading: GB/T18781-2008 [S]. Beijing,China: Standards Press of China, 2008 (in Chinese).
[7]	BLASCO J, CUBERO S, G′OMEZ-SANCH′IS J, et al.Development of a machine for the automatic sorting ofpomegranate (punica granatum) arils based on computervision [J]. Journal of Food Engineering, 2009,90(1): 27-34.
[8]	LIU C Q, CHEN B Q. Method of image detectionfor ear of corn based on computer vision [J]. Transactionsof the Chinese Society of Agricultural Engineering,2014, 30(6): 131-138 (in Chinese).
[17]	ERIKSSON A, PHAM T T, CHIN T J, et al. The ksupportnorm and convex envelopes of cardinality andrank [C]//2015 IEEE Conference on Computer Visionand Pattern Recognition (CVPR). Boston, USA:IEEE, 2015: 3349-3357.
[9]	YANG Z Y, ZHANG W Q, LI W, et al. Informationacquisition method of potted-seedling transplantingfitness using monocular vision [J]. Transactions ofthe Chinese Society of Agricultural Engineering, 2014,30(3): 112-119 (in Chinese).
[18]	RUPPRECHT C, PETER L, NAVAB N. Image segmentationin twenty questions [C]//2015 IEEE Conferenceon Computer Vision and Pattern Recognition(CVPR). Boston, USA: IEEE, 2015: 3314-3322.
[10]	ZHOU Z, HUANG Y, LI X Y, et al. Automatic detectingand grading method of potatoes based on machinevision [J]. Transactions of the Chinese Societyof Agricultural Engineering, 2012, 28(7): 178-183 (inChinese).
[19]	LU C W, LIU S, JIA J Y, et al. Contour box: Rejectingobject proposals without explicit closed contours[C]//2015 IEEE International Conference on ComputerVision (ICCV ). Santiago, Chile: IEEE, 2015:2021-2029.
[11]	LI G, LI B, WANG Y, et al. Pearl shape recognitionbased on computer vision [J]. Transactions of the ChineseSociety for Agricultural Machinery, 2008, 39(7):129-132 (in Chinese).
[20]	DUAN Y F, WANG Q H, LI X M, et al. Highthroughputonline detection method of egg size andshape based on convex hull algorithm [J]. Transactionsof the Chinese Society of Agricultural Engineering,2016, 32(15): 282-288 (in Chinese).
[21]	LIU G L, GINGOLD Y, LIEN J M. Continuous visibilityfeature [C]//2015 IEEE Conference on ComputerVision and Pattern Recognition (CVPR). Boston,USA: IEEE, 2015: 1182-1190.
[12]	TANG Y P, XIA S J, FENG Y J, et al. Pearl onlinedetecting and grading device based on monocularmulti-view machine vision [J]. Transactions of the ChineseSociety for Agricultural Machinery, 2014, 45(1):288-292 (in Chinese).
[13]	WANG YW, QU G T, LIU X L, et al. Image subtractiondetection algorithm for surface defect of steel ball[J]. Journal of Computer-Aided Design & ComputerGraphics, 2016, 28(10): 1699-1704 (in Chinese).
[22]	OPELT A, PINZ A, ZISSERMAN A. Learning an alphabetof shape and appearance for multi-class objectdetection [J]. International Journal of Computer Vision,2008, 80(1): 16-44.
[14]	SUN H, WANG Z, FU L H, et al. Analysis andimplementation of novel method for steel ball surfacedetection [J]. Mechanical Science and Technologyfor Aerospace Engineering, 2016, 35(1): 118-121 (inChinese).
[23]	GIACHETTI A, LOVATO C. Radial symmetry detectionand shape characterization with the multiscalearea projection transform [J]. Computer Graphics Forum,2012, 31(5): 1669-1678.
[24]	ZHAO L K, SONG W D, WANG J X. Straight line extractionalgorithm of freeman chain code priority [J].Geomatics and Information Science of Wuhan University,2014, 39(1): 42-46 (in Chinese).
[15]	QUAN Y M, ZHU G Q, JIANG C C, et al. Makingsmall bell welter for automatic detection of surface defect[J]. Modern Manufacturing Engineering, 2010 (9):115-117 (in Chinese).
[25]	FREEMAN H. Computer processing of line-drawingimages [J]. Computer Surveys, 1974, 6(1): 57-97.
[16]	Standardization Administration of the P.R.C. Culturedpearl grading: GB/T18781-2008 [S]. Beijing,China: Standards Press of China, 2008 (in Chinese).
[26]	DONG Z P, HUANG F G. Review of measurementand evaluation methods for roundness error [J]. ToolEngineering, 2011, 45(2): 14-19 (in Chinese).
[27]	FATEMI M, AMINI A, BABOULAZ L, et al. Shapesfrom pixels [J]. IEEE Transactions on Image Processing,2015, 25(3): 1193-1206.
[17]	ERIKSSON A, PHAM T T, CHIN T J, et al. The ksupportnorm and convex envelopes of cardinality andrank [C]//2015 IEEE Conference on Computer Visionand Pattern Recognition (CVPR). Boston, USA:IEEE, 2015: 3349-3357.
[28]	OTSU N. A thresholding selection method from graylevelhistogram [J]. IEEE Transactions on Systems,Man, and Cybernetics, 1979, 9(1): 62-66.
[18]	RUPPRECHT C, PETER L, NAVAB N. Image segmentationin twenty questions [C]//2015 IEEE Conferenceon Computer Vision and Pattern Recognition(CVPR). Boston, USA: IEEE, 2015: 3314-3322.
[29]	WEINZAEPFEL P, REVAUD J, HARCHAOUI Z, etal. Learning to detect motion boundaries [C]//2015IEEE Conference on Computer Vision and PatternRecognition (CVPR). Boston, USA: IEEE, 2015: 2578-2586.
[19]	LU C W, LIU S, JIA J Y, et al. Contour box: Rejectingobject proposals without explicit closed contours[C]//2015 IEEE International Conference on ComputerVision (ICCV ). Santiago, Chile: IEEE, 2015:2021-2029.
[20]	DUAN Y F, WANG Q H, LI X M, et al. Highthroughputonline detection method of egg size andshape based on convex hull algorithm [J]. Transactionsof the Chinese Society of Agricultural Engineering,2016, 32(15): 282-288 (in Chinese).
[21]	LIU G L, GINGOLD Y, LIEN J M. Continuous visibilityfeature [C]//2015 IEEE Conference on ComputerVision and Pattern Recognition (CVPR). Boston,USA: IEEE, 2015: 1182-1190.
[22]	OPELT A, PINZ A, ZISSERMAN A. Learning an alphabetof shape and appearance for multi-class objectdetection [J]. International Journal of Computer Vision,2008, 80(1): 16-44.
[23]	GIACHETTI A, LOVATO C. Radial symmetry detectionand shape characterization with the multiscalearea projection transform [J]. Computer Graphics Forum,2012, 31(5): 1669-1678.
[24]	ZHAO L K, SONG W D, WANG J X. Straight line extractionalgorithm of freeman chain code priority [J].Geomatics and Information Science of Wuhan University,2014, 39(1): 42-46 (in Chinese).
[25]	FREEMAN H. Computer processing of line-drawingimages [J]. Computer Surveys, 1974, 6(1): 57-97.
[26]	DONG Z P, HUANG F G. Review of measurementand evaluation methods for roundness error [J]. ToolEngineering, 2011, 45(2): 14-19 (in Chinese).
[27]	FATEMI M, AMINI A, BABOULAZ L, et al. Shapesfrom pixels [J]. IEEE Transactions on Image Processing,2015, 25(3): 1193-1206.
[28]	OTSU N. A thresholding selection method from graylevelhistogram [J]. IEEE Transactions on Systems,Man, and Cybernetics, 1979, 9(1): 62-66.
[29]	WEINZAEPFEL P, REVAUD J, HARCHAOUI Z, etal. Learning to detect motion boundaries [C]//2015IEEE Conference on Computer Vision and PatternRecognition (CVPR). Boston, USA: IEEE, 2015: 2578-2586.