CN113504238B - Glass surface defect acquisition device and detection method - Google Patents
Glass surface defect acquisition device and detection method Download PDFInfo
- Publication number
- CN113504238B CN113504238B CN202110627103.0A CN202110627103A CN113504238B CN 113504238 B CN113504238 B CN 113504238B CN 202110627103 A CN202110627103 A CN 202110627103A CN 113504238 B CN113504238 B CN 113504238B
- Authority
- CN
- China
- Prior art keywords
- defect
- light source
- box body
- glass
- conveyor belt
- 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.)
- Active
Links
- 230000007547 defect Effects 0.000 title claims abstract description 104
- 239000011521 glass Substances 0.000 title claims abstract description 57
- 238000001514 detection method Methods 0.000 title claims abstract description 26
- 238000000034 method Methods 0.000 claims description 18
- 238000012549 training Methods 0.000 claims description 17
- 238000000605 extraction Methods 0.000 claims description 8
- 238000004519 manufacturing process Methods 0.000 claims description 8
- 238000012360 testing method Methods 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 5
- 238000000137 annealing Methods 0.000 claims description 4
- 238000002790 cross-validation Methods 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 4
- 238000013519 translation Methods 0.000 claims description 4
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 238000013135 deep learning Methods 0.000 claims description 3
- 230000002708 enhancing effect Effects 0.000 claims description 3
- 238000005728 strengthening Methods 0.000 claims description 3
- 238000010200 validation analysis Methods 0.000 claims description 3
- 238000013136 deep learning model Methods 0.000 abstract description 3
- 239000002699 waste material Substances 0.000 description 4
- 239000000428 dust Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/01—Arrangements or apparatus for facilitating the optical investigation
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/95—Investigating the presence of flaws or contamination characterised by the material or shape of the object to be examined
- G01N21/958—Inspecting transparent materials or objects, e.g. windscreens
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration using histogram techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/01—Arrangements or apparatus for facilitating the optical investigation
- G01N2021/0106—General arrangement of respective parts
- G01N2021/0112—Apparatus in one mechanical, optical or electronic block
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
- G01N2021/8854—Grading and classifying of flaws
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
- G01N2021/8887—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges based on image processing techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
- G06T2207/20032—Median filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Pathology (AREA)
- Biochemistry (AREA)
- Immunology (AREA)
- Data Mining & Analysis (AREA)
- Analytical Chemistry (AREA)
- Chemical & Material Sciences (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Signal Processing (AREA)
- Quality & Reliability (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The invention discloses a glass surface defect acquisition device and a detection method, wherein the acquisition device comprises a box body, an industrial camera, an annular light source and a conveyor belt; the industrial camera is arranged at the top end of the inside of the box body, the conveyor belt is arranged at the bottom end of the inside of the box body, and the annular light source is arranged between the industrial camera and the conveyor belt; the glass is placed on a conveyor. The invention can provide irradiation conditions with good uniformity, good stability and high brightness by using the low-angle annular light source to collect the defect information of the planar glass, can furthest improve the acquisition of the defect information, and particularly can reduce misjudgment and missed judgment aiming at the defects of relatively tiny scratch depth and difficult recognition, thereby greatly improving the accuracy of a detection system; and the deep learning model is used for identifying and classifying the defect information, so that the defect information on the curved surface can be effectively collected, and the defect type can be rapidly and efficiently detected.
Description
Technical Field
The invention relates to the technical field of defect detection, in particular to a glass surface defect acquisition device and a glass surface defect detection method.
Background
Glass is an important basic industrial raw material for national economy, and various defects can occur in a complex deep processing process under the influence of technical conditions, production environments or human factors. The defects on the surface of the glass destroy the appearance quality and optical uniformity of a glass product, reduce the use value of the glass, and further aggravate the waste of resources if the defects generated in the previous process flow into the next process, so that the glass needs to be detected in time in the production process. The existing glass surface defect detection method is mainly based on a traditional machine vision mode, but the traditional machine vision needs to manually extract the characteristics of the region of interest, and has the disadvantages of long time consumption, high difficulty and non-uniform standard, so that a great deal of waste of human resources is caused; and the defects of the glass are numerous and diversified, so that the recognition rate is low, the defect detection speed is low, and the efficiency is low.
Disclosure of Invention
Therefore, the technical problem to be solved by the invention is to overcome the defects in the prior art, so as to provide a glass surface defect acquisition device and a glass surface defect detection method.
For this purpose, the glass surface defect acquisition device comprises a box body, an industrial camera, an annular light source and a conveyor belt; the industrial camera is arranged at the top end of the inside of the box body, the conveyor belt is arranged at the bottom end of the inside of the box body, and the annular light source is arranged between the industrial camera and the conveyor belt; the glass is placed on a conveyor.
Preferably, the light source device further comprises a light source bracket, wherein the light source bracket is arranged on the box body and can move up and down in the box body; the annular light source is arranged on the light source bracket.
Preferably, the camera fixture is mounted at the top end inside the box body, and the industrial camera is mounted inside the camera fixture.
Preferably, the box body further comprises a base, and the box body is mounted on the base.
Preferably, the conveyor belt further comprises a driving motor, and the rotation of the conveyor belt is controlled by the driving motor.
The invention also provides a glass surface defect detection method by using the glass surface defect acquisition device, which comprises the following steps:
s1: the height of the annular light source is adjusted to the height of the maximum width of the glass plane which can be collected by the industrial camera, and pictures are collected;
s2: extracting a circular area polished by an annular light source by adopting an image processing technology, removing redundant interference information, and extracting an ROI (region of interest) area;
s3: extracting the characteristics of the ROI region of interest after the extraction is finished, and strengthening the characteristic information;
s4: using label software to frame and classify targets such as scratches, dirt, edge breakage and the like on the surface of the glass, and manufacturing a data set;
s5: selecting YOLOv4 as a basic algorithm frame, constructing a YOLOv4 deep learning defect detection network, optimizing a training strategy, learning defect information through a neural network according to defect information marked by a training set by the model, obtaining a training weight model, and using the training weight model in a server defect detection module.
Preferably, step S2 includes:
s21: carrying out image enhancement on the image, and enhancing the characteristic information of the defect;
s22: denoising the image by adopting median filtering;
s23: dividing regions by adopting a gray level histogram, selecting a circular region needing defect information identification, and eliminating other interference regions.
Preferably, step S3 includes:
s31: performing edge extraction on the defect information by using a sobel operator, and highlighting the defect information;
s32: fitting according to the shape of the defect information to finish part of defect information;
s33: the number of data samples is amplified by performing a rotation change, a scale change and a translation change on the image.
Preferably, step S4 includes:
s41: dividing defects into three types, namely scratches (flaws, points, foreign matters), dirt and broken edges, and manufacturing a peer positive sample set, namely a planar glass sample without defects;
s42: selecting defect characteristics of a required detected target by using a label tag software frame, generating a corresponding xml file, wherein the xml file contains the position and category information of the defect characteristics selected by the frame;
s43: the dataset was divided into three parts, namely 80% training set, 10% test set and 10% validation set.
Preferably, the method further comprises: adopting a DropBlock regularization method, mosaic data enhancement and cosine annealing learning rate to further optimize the model; and the accuracy of the test result is improved by adopting a K-fold cross validation method.
According to the glass surface defect acquisition device and the glass surface defect detection method, the low-angle annular light source is used for acquiring the defect information of the planar glass, so that irradiation conditions with good uniformity, good stability and high brightness can be provided, the acquisition of the defect information can be improved to the greatest extent, and especially, for the defects which are tiny, shallow in scratch depth and difficult to identify, misjudgment and missed judgment can be reduced, and the accuracy of a detection system is improved to the greatest extent; the deep learning model is used for identifying and classifying the defect information, the characteristics of the region of interest do not need to be manually extracted, the time consumption is short, the difficulty is low, a great deal of waste of human resources can be avoided, the defect information on the curved surface can be effectively collected, and the defect type can be rapidly and efficiently detected.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the description of the embodiments or the prior art will be briefly described, and it is obvious that the drawings in the description below are some embodiments of the present invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
Fig. 1 is a schematic diagram of a glass surface defect collecting device provided by the invention.
Reference numerals: 1. a case; 2. an industrial camera; 3. an annular light source; 4. a conveyor belt; 5. a light source support; 6. a camera fixture; 7. a base; 8. and driving the motor.
Detailed Description
The following description of the embodiments of the present invention will be made apparent and fully in view of the accompanying drawings, in which some, but not all embodiments of the invention are shown. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. are based on the directions or positional relationships shown in the drawings, are merely for convenience of describing the present invention and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a specific orientation, be configured and operated in a specific orientation, and thus should not be construed as limiting the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
In the description of the present invention, it should be noted that, unless explicitly specified and limited otherwise, the terms "mounted," "connected," and "connected" are to be construed broadly, and may be either fixedly connected, detachably connected, or integrally connected, for example; can be mechanically or electrically connected; can be directly connected or indirectly connected through an intermediate medium, and can be communication between two elements. The specific meaning of the above terms in the present invention will be understood in specific cases by those of ordinary skill in the art.
In addition, the technical features of the different embodiments of the present invention described below may be combined with each other as long as they do not collide with each other.
As shown in fig. 1, the present embodiment provides a glass surface defect collecting device, which includes a case 1, an industrial camera 2, an annular light source 3, and a conveyor belt 4; the industrial camera 2 is arranged at the top end inside the box body 1, the conveyor belt 4 is arranged at the bottom end inside the box body 1, and the annular light source 3 is arranged between the industrial camera 2 and the conveyor belt 4; the glass is placed on a conveyor belt 4.
In this embodiment, since dust in the air interferes with the quality of the collected image under the irradiation of strong light, the dust is easily identified as a defect, and a judgment error is caused, the collection device is operated in a closed space by arranging the box 1, so that the probability of erroneous judgment can be reduced.
In this embodiment, the annular light source 3 can provide illumination conditions with good uniformity, good stability and high brightness, can furthest improve the acquisition of defect information, and especially can reduce misjudgment and missed judgment aiming at the defects of relatively tiny scratch depth and difficult identification, thereby greatly improving the accuracy of the detection system.
The glass surface defect acquisition device further comprises a light source bracket 5, wherein the light source bracket 5 is arranged on the box body 1 and can move up and down in the box body 1; the annular light source 3 is mounted on a light source holder 5.
In the present embodiment, since the effect produced by the annular light source 3 is circular, information of the glass surface in the range of the light source is not displayed as the whole glass surface. By mounting the annular light source 3 on the light source support 5, the height of the light source support 5 and thus the annular light source 3 can be adjusted, so that the annular light source 3 can cover the whole glass, and the industrial camera 2 can collect the whole glass plane.
The glass surface defect acquisition device further comprises a camera clamp 6, the camera clamp 6 is arranged at the top end inside the box body 1, and the industrial camera 2 is arranged in the camera clamp 6.
In the present embodiment, the camera jig 6 is used for mounting the industrial camera 2, and by mounting the industrial camera 2 in the camera jig 6, it is possible to ensure that the industrial camera 2 can be stably mounted on the inside top end of the case 1.
The glass surface defect acquisition device further comprises a base 7, and the box body 1 is arranged on the base 7; the glass surface defect acquisition device further comprises a driving motor 8, and the conveyor belt 4 is controlled to rotate by the driving motor 8.
In this embodiment, the base 7 is used for fixing the case 1, and the driving motor 8 is used for driving the conveyor belt 4 to rotate.
In this embodiment, the process of collecting the glass surface defect image is as follows: the driving motor 8 is controlled to move, the conveyor belt 4 starts to move from right to left, the plane glass moves from right to left, in the moving process, the industrial camera 2 collects images on the surface of the glass from left to right, the light source support 5 is fixed at a certain height of the box body 1, the height of the light source support 5 is controlled by the linear module of the z-axis, the annular light source 3 is fixed on the light source support 5, and the brightness of the annular light source 3 is adjusted by the power supply controller.
The embodiment also provides a glass surface defect detection method using the glass surface defect acquisition device, which comprises the following steps:
s1: the height of the annular light source 3 is adjusted to the height of the industrial camera 2 capable of collecting the maximum width of the glass plane, and pictures are collected;
s2: extracting a circular area polished by the annular light source 3 by adopting an image processing technology, removing redundant interference information, and extracting an ROI interest area;
s3: extracting the characteristics of the ROI region of interest after the extraction is finished, and strengthening the characteristic information;
s4: using label software to frame and classify targets such as scratches, dirt, edge breakage and the like on the surface of the glass, and manufacturing a data set;
s5: selecting YOLOv4 as a basic algorithm frame, constructing a YOLOv4 deep learning defect detection network, optimizing a training strategy, learning defect information through a neural network according to defect information marked by a training set by the model, obtaining a training weight model, and using the training weight model in a server defect detection module.
In this embodiment, since the industrial camera 2 needs to be adjusted to a certain height to cause that the range of the identified area is too large compared with the detection effective area, after the picture is collected, the circular area polished by the annular light source 3 is extracted by adopting an image processing technology, redundant interference information is removed, and after the extraction of the identified area is completed, the characteristic extraction is performed on the identified area, so that the characteristic information is enhanced.
The step S2 comprises the following steps:
s21: carrying out image enhancement on the image, and enhancing the characteristic information of the defect;
s22: denoising the image by adopting median filtering;
s23: dividing regions by adopting a gray level histogram, selecting a circular region needing defect information identification, and eliminating other interference regions.
In the present embodiment, since noise is unavoidable in the digital-to-electrical conversion process of the industrial camera 2, median filtering is used to denoise the image, and sharp edge information of the image can be protected.
The step S3 comprises the following steps:
s31: performing edge extraction on the defect information by using a sobel operator, and highlighting the defect information;
s32: fitting according to the shape of the defect information to finish part of defect information;
s33: the number of data samples is amplified by performing a rotation change, a scale change and a translation change on the image.
In the present embodiment, step S2 and step S3 are image preprocessing. Because part of the defect has part of information lost in the image preprocessing process, the defect information is discontinuous, and the part of defect information needs to be fitted according to the shape of the defect information to finish the part of defect information. Since the number of defective samples is small, it is necessary to amplify the number of data samples by performing operations such as rotation change, scale change and translation change on the image.
The step S4 includes:
s41: dividing defects into three types, namely scratches (flaws, points, foreign matters), dirt and broken edges, and manufacturing a peer positive sample set, namely a planar glass sample without defects;
s42: selecting defect characteristics of a required detected target by using a label tag software frame, generating a corresponding xml file, wherein the xml file contains the position and category information of the defect characteristics selected by the frame;
s43: the dataset was divided into three parts, namely 80% training set, 10% test set and 10% validation set.
The glass surface defect detection method further comprises the following steps: adopting a DropBlock regularization method, mosaic data enhancement and cosine annealing learning rate to further optimize the model; and the accuracy of the test result is improved by adopting a K-fold cross validation method.
In this embodiment, the model structure used is a YOLOv4 network, and the size of the input of the network is a multiple of 32, and a proper size is selected according to the performance of the GPU of the computer. In the selection of the industrial camera 2, a 500-ten thousand-pixel camera is used, which is sufficient in resolution, and the final selection of the input picture size is 608 x 608. In order to further optimize the model, a DropBlock regularization method, a Mosaic data enhancement method, a cosine annealing learning rate and the like are adopted, and in order to improve the accuracy of a test result, a K-fold cross validation method is adopted for verification. After the model is trained, training weights are obtained, and the training weights are loaded into a prediction program to detect whether the glass plane has defects or not.
According to the glass surface defect acquisition device and the glass surface defect detection method, the low-angle annular light source is used for acquiring the defect information of the planar glass, so that irradiation conditions with good uniformity, good stability and high brightness can be provided, the acquisition of the defect information can be improved to the greatest extent, and especially, for the defects which are tiny, shallow in scratch depth and difficult to identify, misjudgment and missed judgment can be reduced, and the accuracy of a detection system is improved to the greatest extent; the deep learning model is used for identifying and classifying the defect information, the characteristics of the region of interest do not need to be manually extracted, the time consumption is short, the difficulty is low, a great deal of waste of human resources can be avoided, the defect information on the curved surface can be effectively collected, and the defect type can be rapidly and efficiently detected.
It is apparent that the above examples are given by way of illustration only and are not limiting of the embodiments. Other variations or modifications of the above teachings will be apparent to those of ordinary skill in the art. It is not necessary here nor is it exhaustive of all embodiments. While still being apparent from variations or modifications that may be made by those skilled in the art are within the scope of the invention.
Claims (1)
1. The glass surface defect detection method is characterized by comprising a glass surface defect acquisition device, wherein the glass surface defect acquisition device comprises the following components: the device comprises a box body (1), an industrial camera (2), an annular light source (3) and a conveyor belt (4); the industrial camera (2) is arranged at the top end inside the box body (1), the conveyor belt (4) is arranged at the bottom end inside the box body (1), and the annular light source (3) is arranged between the industrial camera (2) and the conveyor belt (4); the glass is placed on a conveyor belt (4);
the light source bracket (5) is arranged on the box body (1) and can move up and down in the box body (1); the annular light source (3) is arranged on the light source bracket (5);
the industrial camera (2) is arranged in the camera clamp (6);
the box body (1) is arranged on the base (7);
the conveyor belt (4) is controlled to rotate by the driving motor (8);
the glass surface defect detection method is characterized by comprising the following steps:
s1: the height of the annular light source (3) is adjusted to the height of the industrial camera (2) capable of collecting the maximum width of the glass plane, and pictures are collected;
s2: extracting a circular area polished by the annular light source (3) by adopting an image processing technology, removing redundant interference information, and extracting an ROI (region of interest) area;
s3: extracting the characteristics of the ROI region of interest after the extraction is finished, and strengthening the characteristic information;
s4: using label software to frame and classify scratch, dirt and edge breakage targets on the surface of the glass, and manufacturing a data set;
s5: selecting YOLOv4 as a basic algorithm frame, constructing a YOLOv4 deep learning defect detection network, optimizing a training strategy, learning defect information by a neural network according to defect information marked by a training set by the model to obtain a training weight model, and using the training weight model in a server defect detection module;
the step S2 comprises the following steps:
s21: carrying out image enhancement on the image, and enhancing the characteristic information of the defect;
s22: denoising the image by adopting median filtering;
s23: dividing areas by adopting a gray level histogram, selecting a circular area needing defect information identification, and eliminating other interference areas;
the step S3 comprises the following steps:
s31: performing edge extraction on the defect information by using a sobel operator, and highlighting the defect information;
s32: fitting according to the shape of the defect information to finish part of defect information;
s33: amplifying the number of data samples by performing rotation change, scale conversion and translation conversion on the image;
the step S4 includes:
s41: dividing defects into three types, namely scratches, dirt and edge breakage, and manufacturing a positive sample set in a peer-to-peer manner;
s42: selecting defect characteristics of a required detected target by using a label tag software frame, generating a corresponding xml file, wherein the xml file contains the position and category information of the defect characteristics selected by the frame;
s43: the dataset was divided into three parts, 80% training set, 10% test set and 10% validation set;
the method further comprises the steps of: adopting a DropBlock regularization method, mosaic data enhancement and cosine annealing learning rate to further optimize the model; and the accuracy of the test result is improved by adopting a K-fold cross validation method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110627103.0A CN113504238B (en) | 2021-06-04 | 2021-06-04 | Glass surface defect acquisition device and detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110627103.0A CN113504238B (en) | 2021-06-04 | 2021-06-04 | Glass surface defect acquisition device and detection method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113504238A CN113504238A (en) | 2021-10-15 |
| CN113504238B true CN113504238B (en) | 2023-12-22 |
Family
ID=78009039
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110627103.0A Active CN113504238B (en) | 2021-06-04 | 2021-06-04 | Glass surface defect acquisition device and detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113504238B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114235847A (en) * | 2021-12-20 | 2022-03-25 | 深圳市尊绅投资有限公司 | Liquid crystal display panel glass substrate surface defect detection device and detection method |
| CN117517348B (en) * | 2023-11-14 | 2024-05-14 | 四川领先微晶玻璃有限公司 | Surface defect detection system based on microcrystalline glass panel finished product |
| CN117890304A (en) * | 2024-03-15 | 2024-04-16 | 内蒙古丰洲材料有限公司 | Steel plate surface defect detection system |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103048333A (en) * | 2012-12-07 | 2013-04-17 | 北京优纳科技有限公司 | Appearance detection equipment and method |
| KR20190114384A (en) * | 2018-03-30 | 2019-10-10 | 광운대학교 산학협력단 | Apparatus and method for skin lesion diagnosis based on neural network |
| CN110389127A (en) * | 2019-07-03 | 2019-10-29 | 浙江大学 | A kind of identification of cermet part and surface defects detection system and method |
| CN110570393A (en) * | 2019-07-31 | 2019-12-13 | 华南理工大学 | mobile phone glass cover plate window area defect detection method based on machine vision |
| CN110672617A (en) * | 2019-09-14 | 2020-01-10 | 华南理工大学 | Method for detecting defects of silk-screen area of glass cover plate of smart phone based on machine vision |
| CN110865081A (en) * | 2019-11-19 | 2020-03-06 | 东莞市翔飞智能装备科技有限公司 | Transparent bottle automatic check out system |
| CN112200762A (en) * | 2020-07-02 | 2021-01-08 | 西南科技大学 | Diode glass bulb defect detection method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101348904B1 (en) * | 2012-01-20 | 2014-01-09 | 한국과학기술원 | Image partitioning method using high-order correlation clustering, system processing the method and recording medium |
-
2021
- 2021-06-04 CN CN202110627103.0A patent/CN113504238B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103048333A (en) * | 2012-12-07 | 2013-04-17 | 北京优纳科技有限公司 | Appearance detection equipment and method |
| KR20190114384A (en) * | 2018-03-30 | 2019-10-10 | 광운대학교 산학협력단 | Apparatus and method for skin lesion diagnosis based on neural network |
| CN110389127A (en) * | 2019-07-03 | 2019-10-29 | 浙江大学 | A kind of identification of cermet part and surface defects detection system and method |
| CN110570393A (en) * | 2019-07-31 | 2019-12-13 | 华南理工大学 | mobile phone glass cover plate window area defect detection method based on machine vision |
| CN110672617A (en) * | 2019-09-14 | 2020-01-10 | 华南理工大学 | Method for detecting defects of silk-screen area of glass cover plate of smart phone based on machine vision |
| CN110865081A (en) * | 2019-11-19 | 2020-03-06 | 东莞市翔飞智能装备科技有限公司 | Transparent bottle automatic check out system |
| CN112200762A (en) * | 2020-07-02 | 2021-01-08 | 西南科技大学 | Diode glass bulb defect detection method |
Non-Patent Citations (3)
| Title |
|---|
| 基于Mask R-CNN的铸件X射线DR图像缺陷检测研究;蔡彪;沈宽;付金磊;张理泽;;仪器仪表学报(第03期);全文 * |
| 基于模糊C均值聚类和Canny算子的红外图像边缘识别与缺陷定量检测;唐庆菊;刘俊岩;王扬;刘元林;梅晨;;红外与激光工程(第09期);全文 * |
| 未成熟芒果的改进YOLOv2识别方法;薛月菊;黄宁;涂淑琴;毛亮;杨阿庆;朱勋沐;杨晓帆;陈鹏飞;;农业工程学报(第07期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113504238A (en) | 2021-10-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113504238B (en) | Glass surface defect acquisition device and detection method | |
| CN111815630B (en) | Defect detection method and device for LCD screen | |
| CN110570393B (en) | Mobile phone glass cover plate window area defect detection method based on machine vision | |
| CN109816644B (en) | Bearing defect automatic detection system based on multi-angle light source image | |
| CN108918543B (en) | Dynamic detection device and method for mirror surface scratch | |
| JP5920994B2 (en) | Method and system for identifying well wall boundaries of microplates | |
| CN112330598B (en) | Method, device and storage medium for detecting stiff yarn defects on chemical fiber surface | |
| CN110445921A (en) | A kind of Mobile phone screen backlight foreign matter defect diagnostic method and device based on machine vision | |
| CN111612737B (en) | Artificial board surface flaw detection device and detection method | |
| WO2012117647A1 (en) | Observation program and observation device | |
| CN101995223A (en) | Chip appearance detection method and system | |
| CN110728681B (en) | Mura defect detection method and device | |
| CN109839384B (en) | Visual detector and detection method for detecting defects of micro vibration motor | |
| JP2013526717A5 (en) | ||
| CN110021012B (en) | Mobile phone lens window glass defect detection method based on machine vision technology | |
| CN106780464A (en) | A kind of fabric defect detection method based on improvement Threshold segmentation | |
| CN109724994A (en) | FPC golden finger surface defects detection system and method based on AOI technology | |
| CN108986055B (en) | Visual detection method for tiny cracks on egg shell surface | |
| CN111474179A (en) | Lens surface cleanliness detection device and method | |
| CN116228651A (en) | Cloth defect detection method, system, equipment and medium | |
| CN113390882A (en) | Tire inner side defect detector based on machine vision and deep learning algorithm | |
| CN114820475B (en) | Edge identification method and system, wafer processing device and method for determining concentric state of wafer and processing table | |
| CN117309892B (en) | Defect detection method, device and system for blue film of battery and light source controller | |
| CN114688969A (en) | Optical lens size detection device and method | |
| CN117252861A (en) | Method, device and system for detecting wafer surface defects |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |