CN107169957A - A kind of glass flaws on-line detecting system and method based on machine vision - Google Patents

A kind of glass flaws on-line detecting system and method based on machine vision Download PDF

Info

Publication number
CN107169957A
CN107169957A CN201710295728.5A CN201710295728A CN107169957A CN 107169957 A CN107169957 A CN 107169957A CN 201710295728 A CN201710295728 A CN 201710295728A CN 107169957 A CN107169957 A CN 107169957A
Authority
CN
China
Prior art keywords
glass
image
feature
image information
flaws
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201710295728.5A
Other languages
Chinese (zh)
Inventor
童凯
董玉婷
宛丽娟
党鹏
王福成
童中凯
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Yanshan University
Original Assignee
Yanshan University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Yanshan University filed Critical Yanshan University
Priority to CN201710295728.5A priority Critical patent/CN107169957A/en
Publication of CN107169957A publication Critical patent/CN107169957A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/0002Inspection of images, e.g. flaw detection
    • G06T7/0004Industrial image inspection
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84Systems specially adapted for particular applications
    • G01N21/88Investigating the presence of flaws or contamination
    • G01N21/95Investigating the presence of flaws or contamination characterised by the material or shape of the object to be examined
    • G01N21/958Inspecting transparent materials or objects, e.g. windscreens
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/10Segmentation; Edge detection
    • G06T7/12Edge-based segmentation
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/10Segmentation; Edge detection
    • G06T7/136Segmentation; Edge detection involving thresholding
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20081Training; Learning
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/30Subject of image; Context of image processing
    • G06T2207/30108Industrial image inspection

Landscapes

  • Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Analytical Chemistry (AREA)
  • General Health & Medical Sciences (AREA)
  • Biochemistry (AREA)
  • Immunology (AREA)
  • Pathology (AREA)
  • Chemical & Material Sciences (AREA)
  • Quality & Reliability (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

The invention discloses a kind of glass flaws on-line detecting system based on machine vision and method, including image capture module, processor module, power module, memory module, display module etc..Image capture module uses linear array CCD camera, progressively scans glass to be measured, obtains glass flaws image information;Core processor module mainly includes the TMS320DM642 DSP minimum systems of TI companies, for receiving and handling glass image information;It mainly completes the pretreatment of image, and carries out feature extraction to flaw image by specific algorithm, calculates its geometric feature and gray feature value, Classification and Identification is carried out to it according to these characteristic parameters;Power module provides the source of energy for whole system.The major function of memory module is to realize the temporary cache of glass view data;Display module mainly realizes the display of testing result image information by USB interface.The present invention improves the real-time of glass flaws on-line checking, enhances reliability.

Description

A kind of glass flaws on-line detecting system and method based on machine vision
Technical field
The present invention relates to glass production manufacture field, the recognition methods of especially a kind of glass flaws based on machine vision And specification processing system.
Background technology
The civilian plate glass of China's production occupy global the first, and glass total growth is probably five points of Gross World Product Two.China has 160 Duo Ge float glass producers at present, and 4/5ths of glass throughput is float glass.According to By advanced glass technology manufacturing technology, the yield and quality of glass increased, but still occur in finished glass Some flaws, influence the performance and outward appearance of glass, such as cut, bubble, tin point, ink dot, light distortion.In the production pin of glass During selling, it is necessary to assure the high-quality of glass, high-performance can just strive to obtain market position, therefore glass flaws detection is glass life Indispensable link in production system, avoids underproof glass from coming into the market with this, more can be to glass unit divisi8 grade. In the past, because technology is limited, the glass production line speed of service is slow, and artificial detection turns into detection method main at that time.It is adjoint The development of science and technology, and the demand of glass is constantly lifted, the production line speed of service of glass is accelerated, artificial detection Limitation highlight, the requirement of production can not be reached.
The shortcoming that artificial detection is present mainly has:Workshop scene dust is more, and noise is big, and workman's detection working environment is disliked It is bad, it is impossible to which that, directly using range estimation, labor intensity is big;Normal eye it is continual observation moving object for a period of time can dim eyesight, Eye is swollen to wait uncomfortable, and testing staff can not non-stop run for a long time, it is impossible to ensure ex factory pass rate;And glass flaws human eye is difficult Accurate to judge, error is big, and error chance is a lot, it is impossible to ensure detection quality, and speed of production also has very big limitation.
Therefore, in glass production processing industry, it is badly in need of one kind and human eye is replaced using electronic system and method to glass Flaw is measured and differentiated.
The content of the invention
Present invention aims at providing, a kind of Detection results are accurate, stability is strong, quickly and easily based on machine vision Glass flaws on-line detecting system and method.
To achieve the above object, following technical scheme is employed:System of the present invention includes linear array CCD camera, embedding Enter formula processor, power module, memory module and display module;
The signal input part connection of linear array CCD camera and embeded processor, linear array CCD camera progressive scan is to be measured Glass, obtains the image information of glass and inputs embeded processor;The signal output part and display module of embeded processor It is connected, embeded processor is separately connected with memory module, power module;Embeded processor is obtained to linear array CCD camera Glass image information pre-processed after, embeded processor by glass information transmit to display module carry out result shows, Glass information is transmitted to memory module and stored by embeded processor;The power module is used to provide a system to electricity Power.
Further, the display module application USB interface, including ADV7123KST140 and peripheral circuit.
Further, the embeded processor is TMS320DM642 processors.
A kind of line detecting method of the present invention, comprises the following steps:
Step 1, linear array CCD camera obtains glass image information and transmitted to embeded processor;
Step 2, embeded processor is pre-processed to glass image information first;Then feature extraction algorithm pair is passed through Glass flaws image carries out feature extraction, calculates its geometric feature and gray feature value, special according to geometric feature and gray scale Value indicative carries out Classification and Identification to glass image information;
Step 3, embeded processor is stored the glass image information write-in memory module after processing, meanwhile, Embeded processor, which transmits the glass image information after processing to display module progress result, to be shown.
Wherein, in step 2, the pretreatment of embeded processor includes filtering and noise reduction, Threshold segmentation processing, in Threshold segmentation After processing, reuse medium filtering and be filtered processing to image, that is, respectively using in once before and after the Threshold segmentation Value filtering carries out denoising, obtains accurate muting flaw bullet image;Comprise the following steps that:
Step 2-1, image processor receives image information, and image median filter, medium filtering are carried out to the image of acquisition Can be with Protect edge information information, computation complexity is smaller, there is good inhibiting effect to salt-pepper noise, therefore it is applied to the glass flaw Defect detecting system.
Step 2-2, after picture noise is removed, to improve recognition efficiency, enters row threshold division.Threshold value is carried out using OTSU Segmentation;
Step 2-3, to the image information obtained after processing, in order to extract accurate flaw core, after threshold process, then Secondary use medium filtering is filtered processing to image, that is, is carried out respectively using a medium filtering before and after Threshold segmentation Denoising, carries out secondary filtering;
Step 2-4, carries out edge extracting processing to glass flaws image using canny operators, makes image boundary clearly demarcated, right Target area identification, extraction, understand that analysis lays the first stone.
In step 2, the feature extraction algorithm, which is used to obtain, is conducive to the feature of defect classification, including Image Moment Invariants The extraction of the characteristics of image such as extraction, the extraction of geometric properties and gray scale, using unbiased U-ReliefF feature selecting algorithms.
In step 3, embeded processor digitally generates image information, and R, G, B three primary colors are changed into by converter Signal and row, field sync signal, signal, into display module, complete the display of testing result image information by cable transmission.
The unbiased U-ReliefF feature selecting algorithms are carried out in feature selecting, and selected five features are:Elongation Degree, rectangular degree, gray standard deviation, threshold value, circularity.
Compared with prior art, the invention has the advantages that:
1st, the real-time of glass flaws on-line checking is improved, reliability is enhanced.
2nd, ReliefF algorithm evaluation accuracys are improved so that it is more fair that feature weight is evaluated, and solves sample size pair The influence of feature weight.
Brief description of the drawings
Fig. 1 is the structured flowchart of present system.
Fig. 2 is the overall workflow figure of present system.
Fig. 3 is image processing program flow chart.
Fig. 4 is image preprocessing flow chart.
Fig. 5 is U-ReliefF algorithm flow charts.
Qualitative description figure of each features of Fig. 6 to flaw.
Fig. 7 characteristic value ordering charts.
Embodiment
The present invention will be further described below in conjunction with the accompanying drawings:
As shown in figure 1, present system is mainly made up of three parts, it is image procossing main control part, IMAQ respectively Part and result display portion.Production line overall construction drawing as indicated with 1, the letter of linear array CCD camera and embeded processor Number input connection, linear array CCD camera progressively scans glass to be measured, and the image information for obtaining glass simultaneously inputs embedded processing Device;The signal output part of embeded processor is connected with display module, embeded processor separately with memory module, power module It is connected;After embeded processor is pre-processed to the glass image information that linear array CCD camera is obtained, embeded processor will Glass information, which is transmitted to display module progress result, to be shown, and glass information is transmitted to memory module and carried out by embeded processor Storage;The power module is used to provide a system to electric power.The system is set based on machine vision by image processing techniques A set of glass flaws detection and identification system are counted, the detecting system obtains flaw image by linear array CCD camera Data, by transmission line data transfer to master control detecting system;When data reach processor, convert the data into first Data signal, and flaw image is pre-processed using image processing techniques, obtain clearly reliably image;Secondly, it is right Flaw image carries out feature extraction, calculates its geometric feature and gray feature, it is divided according to these characteristic parameters Class is recognized.
IMAQ and image processing module in the system, described image acquisition module include SG-14-04K80 lines Array camera, Myutron FV5026L-F camera lenses, the red strip sources of LED of 620nm-750nm wavelength, shone using backlight type Penetrate, and peripheral circuit, pending image is obtained with this.Image processing module using embedded processing systems as core at Device is managed, the system overall workflow built is as shown in Figure 2.
The DSP programmings of image procossing, by linear array CCD camera collection Lai image information be temporarily present the slow of system Rush in memory, and view data is transferred in DSP according to the DSP intrinsic clock cycle.Its internal image processing routine Flow is as shown in Figure 3.Because glass flaws image data amount is big, so system needs extension external mass storage to deposit Storage, so DSP is accomplished by keeping data communication with external memory storage, realizes the real-time reading, processing and transmission of data.
The glass image gathered on image preprocessing process, production line to it, it is necessary to carry out after image preprocessing Feature extraction and flaw identification can effectively be carried out.The process of image preprocessing is mainly completion following items task:Image Filtering, Threshold segmentation, rim detection etc., as shown in Figure 4.
Step one:Image processor receives image information, carries out image median filter to the image of acquisition, medium filtering can With Protect edge information information well, computation complexity is smaller, there is good inhibiting effect to salt-pepper noise, therefore it is applied to glass Glass Defect Detection system.
Step 2:After picture noise is removed, to improve recognition efficiency, enter row threshold division.Threshold value is carried out using OTSU Segmentation.
Step 3:To the image information obtained after processing, in order to extract accurate flaw core, after threshold process, then Secondary use medium filtering is filtered processing to image.Carried out respectively using a medium filtering namely before and after Threshold segmentation Denoising, carries out secondary filtering.
Step 4:Edge extracting processing is carried out to glass flaws image using canny operators, makes image boundary clearly demarcated, it is right Target area identification, extraction, understand that analysis lays the first stone.
Get after glass flaws area-of-interest, target subject is clear and definite, but only extract with good discrimination The feature of degree, computer just meets with figure and knows thing.So-called feature refers to the common property that interesting image regions possess, and is also meter The foundation of calculation machine identification.The extraction of unwanted visual characteristic is that from the object space of image glass flaws are mapped into feature space, this Mapping relations reduce the information content of flaw image, are that follow-up identification classification lays the foundation.
The feature selection approach of suitable glass flaws classification is explored, a kind of feature selecting algorithm model U- of unbiased is proposed ReliefF.Algorithm flow chart such as Fig. 5.
Step one:The feature extraction of flaw image.The geometric properties of glass flaws are the notable features for detecting target.Generally In the case of, geometry is counted as a closing, continuous region, and the profile in this region is a curve, therefore figure As the description of geometric properties is exactly to this closing, continuous region description;These geometric properties of Fig. 6 describe image jointly Overall distribution, therefore avoid the influence that the local feature of image is brought, there is good tolerance.
Step 2:The feature selecting of flaw image.The target of feature selecting is to find optimal feature subset.Feature selecting energy The feature of uncorrelated (irrelevant) or redundancy (redundant) is rejected, so as to reach reduction Characteristic Number, model essence is improved Exactness, reduces the purpose of run time.On the other hand, very positively related feature reduction model is selected, assistance understands that data are produced Raw process.
Step 3:Relief algorithms are probed into.The core concept of the algorithm is:Distance of the good feature between similar sample It is smaller, and the distance between inhomogeneous sample is larger;For a multiclass sample, its inter- object distance is discrete by similar sample Degree influence, the high sample of within-cluster variance, value is at a higher level.And comparing the sample compacted in class, value is in One relatively low level.Tagsort ability is assessed with desired value, obtained result is partial to the sample class being dominant in quantity.
Step 4:The foundation of U-Relief algorithm models.In practical application, it is impossible to ensure that the sample of each feature class to the greatest extent may be used Many, the having differences property of quantity of sample of energy.In the case, to improve ReliefF algorithm evaluation accuracys, its algorithm is entered Row is improved, as shown in Figure 5.
In order to which more popular understanding features described above is extracted and selection step, illustrated with reference to concrete instance:
For actual production scene extracted unwanted visual characteristic (including:Gray standard deviation (1), threshold value (2), circularity (3), rectangular degree (4), elongation (5), gradient (6), eccentricity (7), length-width ratio (8), compactedness (9), range of extension (10) etc. Geometric properties), using U-ReliefF algorithms carry out evaluation selection, while with ReliefF algorithm comparisons, verify innovatory algorithm pair The validity of glass flaws feature selecting.
For proving and comparisom algorithm, the unwanted visual characteristic of said extracted is commented using ReliefF and U-ReliefF algorithms Valency is selected, and the feature weight sequence for calculating each feature is as shown in Figure 7.
By Fig. 7 it can also be seen that ReliefF algorithms are carried out in feature selecting, preceding four characteristic values are relatively stablized, the 5th spy Value indicative increases higher, if using amplification more than 20% as threshold point, the 5th later feature is chosen, that is, is tilted Degree, elongation, rectangular degree, gray standard deviation, threshold value, 6 features of circularity are selected as the foundation of classification;Similarly understand, The feature of innovatory algorithm selection is elongation, gradient, circularity, threshold value, 5 features of gray standard deviation, ReliefF algorithms In rectangular degree feature reject, this is due to that this feature tends to the more tin point of sample in training set, to identification tin point play Positive role, but act on unobvious for general classification, wrong point is even resulted in, therefore rejected in innovatory algorithm.
The feature gone out according to two kinds of algorithms selections, carries out class test to training sample set respectively.It is selected without feature Classification accuracy is 71.96% when selecting, and accuracy rate is 82.83% after ReliefF feature selectings, is calculated using U-ReliefF Method accuracy rate is 85.69%.By feature selecting, total classification recognition accuracy improves 13.73%, wherein relatively being improved after improving Before improve 2.86%, illustrate necessity of the feature selecting to glass flaws Classification and Identification.
Embodiment described above is only that the preferred embodiment of the present invention is described, not to the model of the present invention Enclose and be defined, on the premise of design spirit of the present invention is not departed from, technical side of the those of ordinary skill in the art to the present invention In various modifications and improvement that case is made, the protection domain that claims of the present invention determination all should be fallen into.

Claims (8)

1. a kind of glass flaws on-line detecting system based on machine vision, it is characterised in that:The system is taken the photograph including line array CCD Camera, embeded processor, power module, memory module and display module;
The signal input part connection of linear array CCD camera and embeded processor, linear array CCD camera progressively scans glass to be measured Glass, obtains the image information of glass and inputs embeded processor;The signal output part of embeded processor and display module phase Even, embeded processor is separately connected with memory module, power module;Embeded processor is obtained to linear array CCD camera After glass image information is pre-processed, embeded processor, which transmits glass information to display module progress result, to be shown, embedding Enter formula processor and transmit glass information to memory module to be stored;The power module is used to provide a system to electric power.
2. a kind of glass flaws on-line detecting system based on machine vision according to claim 1, it is characterised in that:Institute State display module application USB interface, including ADV7123KST140 and peripheral circuit.
3. a kind of glass flaws on-line detecting system based on machine vision according to claim 1, it is characterised in that:Institute Embeded processor is stated for TMS320DM642 processors.
4. a kind of glass flaws online test method based on system described in claim 1, it is characterised in that methods described includes Following steps:
Step 1, linear array CCD camera obtains glass image information and transmitted to embeded processor;
Step 2, embeded processor is pre-processed to glass image information first;Then by feature extraction algorithm to glass Flaw image carries out feature extraction, calculates its geometric feature and gray feature value, according to geometric feature and gray feature value Classification and Identification is carried out to glass image information;
Step 3, embeded processor is stored the glass image information write-in memory module after processing, meanwhile, it is embedded Formula processor, which transmits the glass image information after processing to display module progress result, to be shown.
5. a kind of glass flaws online test method based on machine vision according to claim 4, it is characterised in that:Step In rapid 2, the pretreatment of embeded processor includes filtering and noise reduction, Threshold segmentation processing, after Threshold segmentation processing, reuses Medium filtering is filtered processing to image, that is, carries out denoising using a medium filtering respectively before and after Threshold segmentation, Obtain accurate muting flaw bullet image;Comprise the following steps that:
Step 2-1, image processor receives image information, and image median filter is carried out to the image of acquisition;
Step 2-2, after picture noise is removed, row threshold division is entered using OTSU;
Step 2-3, to the image information obtained after processing, after threshold process, reuses medium filtering and image is filtered Ripple processing, that is, denoising is carried out using a medium filtering respectively before and after Threshold segmentation, carry out secondary filtering;
Step 2-4, carries out edge extracting processing to glass flaws image using canny operators, makes image boundary clearly demarcated, to target Region recognition, extraction, understand that analysis lays the first stone.
6. a kind of glass flaws online test method based on machine vision according to claim 4, it is characterised in that:Step In rapid 2, the feature extraction algorithm is used to obtain the feature for being conducive to defect classification, includes extraction, the geometry of Image Moment Invariants The extraction of the characteristics of image such as the extraction of feature and gray scale, using unbiased U-ReliefF feature selecting algorithms.
7. a kind of glass flaws online test method based on machine vision according to claim 4, it is characterised in that:Step In rapid 3, embeded processor digitally generates image information, and R, G, B tricolor signal and row, field are changed into by converter Synchronizing signal, signal, into display module, completes the display of testing result image information by cable transmission.
8. a kind of glass flaws online test method based on machine vision according to claim 6, it is characterised in that:Institute State unbiased U-ReliefF feature selecting algorithms to carry out in feature selecting, selected five features are:Elongation, rectangular degree, ash Spend standard deviation, threshold value, circularity.
CN201710295728.5A 2017-04-28 2017-04-28 A kind of glass flaws on-line detecting system and method based on machine vision Pending CN107169957A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710295728.5A CN107169957A (en) 2017-04-28 2017-04-28 A kind of glass flaws on-line detecting system and method based on machine vision

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710295728.5A CN107169957A (en) 2017-04-28 2017-04-28 A kind of glass flaws on-line detecting system and method based on machine vision

Publications (1)

Publication Number Publication Date
CN107169957A true CN107169957A (en) 2017-09-15

Family

ID=59812416

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710295728.5A Pending CN107169957A (en) 2017-04-28 2017-04-28 A kind of glass flaws on-line detecting system and method based on machine vision

Country Status (1)

Country Link
CN (1) CN107169957A (en)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107742301A (en) * 2017-10-25 2018-02-27 哈尔滨理工大学 Transmission line of electricity image processing method under complex background based on image classification
CN108000239A (en) * 2017-12-01 2018-05-08 常州信息职业技术学院 Digital control processing on-line detecting system
CN108596249A (en) * 2018-04-24 2018-09-28 苏州晓创光电科技有限公司 The method and apparatus of image characteristics extraction and classification
CN110232681A (en) * 2019-05-29 2019-09-13 深圳新视智科技术有限公司 Fault methods of exhibiting, device, computer equipment and storage medium
CN110596126A (en) * 2018-05-25 2019-12-20 上海翌视信息技术有限公司 Sheet glass edge flaw detection method based on image acquisition
CN111223080A (en) * 2020-01-02 2020-06-02 长江存储科技有限责任公司 Wafer detection method and device, electronic equipment and storage medium
CN116542975A (en) * 2023-07-05 2023-08-04 成都数之联科技股份有限公司 Defect classification method, device, equipment and medium for glass panel

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102305798A (en) * 2011-08-02 2012-01-04 上海交通大学 Method for detecting and classifying glass defects based on machine vision
CN105044119A (en) * 2015-08-27 2015-11-11 李明英 Glass flaw classification method based on gray mean value analysis
CN105184783A (en) * 2015-08-27 2015-12-23 袁芬 Glass defect type identification system

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102305798A (en) * 2011-08-02 2012-01-04 上海交通大学 Method for detecting and classifying glass defects based on machine vision
CN105044119A (en) * 2015-08-27 2015-11-11 李明英 Glass flaw classification method based on gray mean value analysis
CN105184783A (en) * 2015-08-27 2015-12-23 袁芬 Glass defect type identification system

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
吴房胜 等: "基于数字图像处理的玻璃瑕疵检测系统", 《宜宾学院学报》 *
王平顺: "图像处理技术在玻璃缺陷检测中的应用研究", 《中国优秀硕士学位论文全文数据库》 *

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107742301A (en) * 2017-10-25 2018-02-27 哈尔滨理工大学 Transmission line of electricity image processing method under complex background based on image classification
CN107742301B (en) * 2017-10-25 2021-07-30 哈尔滨理工大学 Image classification-based power transmission line image processing method under complex background
CN108000239A (en) * 2017-12-01 2018-05-08 常州信息职业技术学院 Digital control processing on-line detecting system
CN108596249A (en) * 2018-04-24 2018-09-28 苏州晓创光电科技有限公司 The method and apparatus of image characteristics extraction and classification
CN108596249B (en) * 2018-04-24 2022-04-22 苏州晓创光电科技有限公司 Image feature extraction and classification method and device
CN110596126A (en) * 2018-05-25 2019-12-20 上海翌视信息技术有限公司 Sheet glass edge flaw detection method based on image acquisition
CN110232681A (en) * 2019-05-29 2019-09-13 深圳新视智科技术有限公司 Fault methods of exhibiting, device, computer equipment and storage medium
CN111223080A (en) * 2020-01-02 2020-06-02 长江存储科技有限责任公司 Wafer detection method and device, electronic equipment and storage medium
CN116542975A (en) * 2023-07-05 2023-08-04 成都数之联科技股份有限公司 Defect classification method, device, equipment and medium for glass panel
CN116542975B (en) * 2023-07-05 2023-09-12 成都数之联科技股份有限公司 Defect classification method, device, equipment and medium for glass panel

Similar Documents

Publication Publication Date Title
CN107169957A (en) A kind of glass flaws on-line detecting system and method based on machine vision
CN106504248B (en) Vehicle damage judging method based on computer vision
CN112818988B (en) Automatic identification reading method and system for pointer instrument
CN110473178A (en) A kind of open defect detection method and system based on multiple light courcess fusion
CN102914545B (en) Gear defect detection method and system based on computer vision
CN104112269B (en) A kind of solar battery laser groove parameter detection method and system based on machine vision
CN103164692B (en) A kind of special vehicle instrument automatic identification system based on computer vision and method
CN108711148B (en) Tire defect intelligent detection method based on deep learning
CN109993094A (en) Fault in material intelligent checking system and method based on machine vision
CN105160652A (en) Handset casing testing apparatus and method based on computer vision
CN102222384A (en) Analysis method of multispectral image of note
CN104063873A (en) Shaft sleeve part surface defect on-line detection method based on compressed sensing
CN106824806A (en) The detecting system and detection method of the low module plastic gear based on machine vision
CN104331978A (en) Recognition device and method for fold of paper currency
CN113252568A (en) Lens surface defect detection method, system, product and terminal based on machine vision
CN101510295B (en) Design method for machine vision system based on PCIe and Vision Assistan
CN113763350A (en) Glue line detection method and device, glue line detection equipment and storage medium
CN204039886U (en) A kind of pavement damage crack detection system based on multiple stage camera stereoscopic shooting
CN102565074B (en) System and method for rechecking images of suspected defective products by small sheet sorter
CN115424115A (en) Transparent target detection method based on polarization imaging and deep learning
CN110175517A (en) A kind of Coal-rock identification method based on transform domain numerical characteristic
CN110009019A (en) Magnetic material crackle intelligent checking system and method
CN117367516B (en) Aluminum-plastic composite board performance detection system based on multidimensional test
CN103424368A (en) Rapid on-site detection method and apparatus for soil salination
CN103179332A (en) Visual target self-adaption detection controller based on field programmable gate array (FPGA) technology

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
WD01 Invention patent application deemed withdrawn after publication
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20170915