CN109975308B - Surface detection method based on deep learning - Google Patents
Surface detection method based on deep learning Download PDFInfo
- Publication number
- CN109975308B CN109975308B CN201910198350.6A CN201910198350A CN109975308B CN 109975308 B CN109975308 B CN 109975308B CN 201910198350 A CN201910198350 A CN 201910198350A CN 109975308 B CN109975308 B CN 109975308B
- Authority
- CN
- China
- Prior art keywords
- image
- target image
- algorithm
- rle
- carrying
- 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
Images
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
- 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
- G06T7/13—Edge detection
-
- 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
-
- 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)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Biochemistry (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Quality & Reliability (AREA)
- Signal Processing (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The invention provides a surface detection method based on deep learning, which comprises the steps of collecting images through a line scanning camera, and then carrying out left and right side edge detection and edge expansion on the collected images to obtain a first target image; carrying out background subtraction algorithm on the first target image to obtain a second target image; carrying out binarization on the second target image to obtain a binary image; performing RLE coding on the binary image to obtain an RLE image; performing particle filtering on the RLE image to obtain the position of a defect particle; and finding the position of the defect particle and carrying out ROI (region of interest) cutting on the region corresponding to the first target image to obtain a cut image, so that the defects and flaws in the web material close to the noise level can be detected.
Description
Technical Field
The invention relates to a surface detection method based on deep learning.
Background
A "web" is a flat material that is continuously produced in large quantities and at very high rates. Common web materials include fabrics, sheet metal, paper, films, nonwovens, and non-woven plastics, among others. During the production process, the surface of the web needs to be inspected for small flaws and defects. The prior art adopts a common binarization technology, can only detect obvious large defects, can not detect small defects and unobvious defects with low contrast, and can not meet the requirements of customers for batch reel materials due to the fact that the defects and the defects can not be detected, thereby causing serious consequences.
Disclosure of Invention
The invention aims to provide a surface detection method based on deep learning, which meets the use requirements of users.
The invention is realized by the following steps: a surface detection method based on deep learning comprises at least one line scan camera and comprises the following steps:
step 1, collecting images through a line scanning camera, and then performing left and right side edge detection and edge expansion on the collected images to obtain a first target image;
step 2, carrying out a background subtraction algorithm on the first target image to obtain a second target image;
step 3, carrying out binarization on the second target image to obtain a binary image;
step 4, RLE coding is carried out on the binary image to obtain an RLE image;
step 5, carrying out particle filtering on the RLE image to obtain the position of a defect particle;
and 6, finding the corresponding area of the first target image for the position of the defective particle, and performing ROI clipping to obtain a clipped image.
Further, the method also comprises a step 7 of accurately classifying the defect particles by deep learning of the cut image by adopting a fast-rcnn algorithm.
Further, the step 1 is further specifically: the method comprises the steps of collecting images through a line scanning camera, then carrying out left and right side edge detection on the collected images by adopting a Canny edge detection algorithm, and then carrying out edge expansion to obtain a first target image.
Further, the step 2 is further specifically: and carrying out a background subtraction algorithm on the first target image by adopting a Gabor filtering algorithm to obtain a second target image.
Further, the step 3 is further specifically: and carrying out binarization on the second target image by using a Niblack binarization algorithm to obtain a binary image.
Further, the step 4 is further specifically: and performing RLE coding on the binary image by adopting an RLE run length coding compression algorithm to obtain an RLE image.
Further, the step 6 is further specifically: and finding a region corresponding to the first target image for the position of the defective particle by adopting a bilinear interpolation algorithm to perform ROI clipping to obtain a clipped image.
The invention has the following advantages: the present invention is connectable to existing factory ethernet web surface inspection systems; the defects can be accurately classified; flaws and defects in the web material can be detected at near noise levels.
Drawings
The invention will be further described with reference to the following examples with reference to the accompanying drawings.
FIG. 1 is a flow chart of the method of the present invention.
Detailed Description
As shown in fig. 1, the present invention is realized by: a surface detection method based on deep learning comprises at least one line scan camera and comprises the following steps:
step 1, collecting images through a line scanning camera, then performing left and right side edge detection on the collected images by adopting a Canny edge detection algorithm, and performing edge expansion to obtain a first target image;
step 2, carrying out a background subtraction algorithm on the first target image by adopting a Gabor filtering algorithm to obtain a second target image;
step 3, carrying out binarization on the second target image by adopting a Niblack binarization algorithm to obtain a binary image;
step 4, RLE coding is carried out on the binary image by adopting an RLE run length coding compression algorithm to obtain an RLE image;
step 5, carrying out particle filtering on the RLE image to obtain the positions of the defective particles;
and 6, finding a region corresponding to the first target image from the position of the defective particle by adopting a bilinear interpolation algorithm, and cutting the ROI to obtain a cut image.
And 7, accurately classifying the defect particles by deep learning of the cut image by adopting a fast-rcnn algorithm.
One embodiment of the present invention:
the roll material detection system at least comprises a line scanning camera for collecting images of roll material parts with flaws or defects, each line scanning camera is connected with at least one PC (personal computer) through a network card, and the PC processes and analyzes the collected images through the surface detection method based on deep learning to obtain information of each flaw and defect, including position and defect type; the PC sends the defect information to the server through the Ethernet and is displayed by a graphical user interface of the server.
In the embodiment of the traditional machine learning and deep learning method, the software programming language adopts C/C + + language, has the advantage of cross-platform, and can adapt to the PC of the current Intel chip and the embedded equipment based on ARM. Has strong universality. The detailed steps are as follows:
step 1: carrying out left and right side edge detection and edge expansion on the collected image to obtain a target image 1; step 2: carrying out background subtraction algorithm on the target image 1 to obtain a target image 2; and step 3: carrying out binarization on the target image 2 to obtain a binary image; and 4, step 4: performing RLE coding on the binary image to obtain an RLE image; and 5: performing particle filtering on the RLE image to obtain the position of a defect particle; step 6: finding a region corresponding to the target image 1 for the position of the defective particle, and performing ROI clipping to obtain a clipped image; and 7: carrying out accurate classification on defective particles of the cut image by using deep learning;
further, step 1, carrying out edge detection and edge expansion on the left side and the right side of the collected image to obtain a target image 1; the implemented algorithm is a Canny edge detection algorithm.
Further, step 2, a background detection algorithm is carried out on the target image 1 to obtain a target image 2, and the background subtraction algorithm is a Gabor filtering algorithm.
Further, step 3, binarization is carried out on the target image 2 to obtain a binary image, and the binarization algorithm is a Niblack binarization algorithm.
Further, step 4, RLE coding is carried out on the binary image to obtain an RLE image, and the RLE algorithm is an RLE run length coding compression algorithm.
And further, step 5, carrying out particle filtering on the RLE image to obtain the position of the defect particle, and carrying out a particle filtering algorithm.
Further, step 6, finding the corresponding area of the image target 1 at the position of the defective particle, and performing ROI clipping to obtain a clipped image, wherein the clipping algorithm is a bilinear interpolation algorithm for clipping.
And step 7, performing accurate defect particle classification on the cut image by using deep learning, wherein the deep learning algorithm is a fast-rcnn algorithm.
Although specific embodiments of the invention have been described above, it will be understood by those skilled in the art that the specific embodiments described are illustrative only and are not limiting upon the scope of the invention, and that equivalent modifications and variations can be made by those skilled in the art without departing from the spirit of the invention, which is to be limited only by the appended claims.
Claims (5)
1. A surface detection method based on deep learning is characterized in that: comprising at least one line-scan camera for acquiring images of portions of web material having flaws or defects, each line-scan camera being connected to at least one PC via a network card, the PC processing the acquired images, comprising the steps of:
step 1, collecting images through a line scanning camera, then performing left and right side edge detection on the collected images by adopting a Canny edge detection algorithm, and performing edge expansion to obtain a first target image;
step 2, carrying out a background subtraction algorithm on the first target image to obtain a second target image;
step 3, carrying out binarization on the second target image to obtain a binary image;
step 4, RLE coding is carried out on the binary image to obtain an RLE image;
step 5, carrying out particle filtering on the RLE image to obtain the positions of the defective particles;
step 6, finding the corresponding area of the first target image for the position of the defect particle, and performing ROI clipping to obtain a clipped image;
and 7, accurately classifying the defect particles by deep learning of the cut image by adopting a fast-rcnn algorithm.
2. The deep learning-based surface detection method according to claim 1, wherein: the step 2 is further specifically as follows: and carrying out a background subtraction algorithm on the first target image by adopting a Gabor filtering algorithm to obtain a second target image.
3. The deep learning-based surface detection method according to claim 1, wherein: the step 3 is further specifically as follows: and carrying out binarization on the second target image by using a Niblack binarization algorithm to obtain a binary image.
4. The deep learning-based surface detection method according to claim 1, wherein: the step 4 is further specifically as follows: and performing RLE coding on the binary image by adopting an RLE run length coding compression algorithm to obtain an RLE image.
5. The deep learning-based surface detection method according to claim 1, wherein: the step 6 is further specifically as follows: and finding a region corresponding to the first target image for the position of the defective particle by adopting a bilinear interpolation algorithm to perform ROI clipping to obtain a clipped image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910198350.6A CN109975308B (en) | 2019-03-15 | 2019-03-15 | Surface detection method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910198350.6A CN109975308B (en) | 2019-03-15 | 2019-03-15 | Surface detection method based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109975308A CN109975308A (en) | 2019-07-05 |
| CN109975308B true CN109975308B (en) | 2023-03-24 |
Family
ID=67079057
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910198350.6A Active CN109975308B (en) | 2019-03-15 | 2019-03-15 | Surface detection method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109975308B (en) |
Families Citing this family (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018176000A1 (en) | 2017-03-23 | 2018-09-27 | DeepScale, Inc. | Data synthesis for autonomous control systems |
| US11157441B2 (en) | 2017-07-24 | 2021-10-26 | Tesla, Inc. | Computational array microprocessor system using non-consecutive data formatting |
| US11409692B2 (en) | 2017-07-24 | 2022-08-09 | Tesla, Inc. | Vector computational unit |
| US11893393B2 (en) | 2017-07-24 | 2024-02-06 | Tesla, Inc. | Computational array microprocessor system with hardware arbiter managing memory requests |
| US10671349B2 (en) | 2017-07-24 | 2020-06-02 | Tesla, Inc. | Accelerated mathematical engine |
| US11561791B2 (en) | 2018-02-01 | 2023-01-24 | Tesla, Inc. | Vector computational unit receiving data elements in parallel from a last row of a computational array |
| US11215999B2 (en) | 2018-06-20 | 2022-01-04 | Tesla, Inc. | Data pipeline and deep learning system for autonomous driving |
| US11361457B2 (en) | 2018-07-20 | 2022-06-14 | Tesla, Inc. | Annotation cross-labeling for autonomous control systems |
| US11636333B2 (en) | 2018-07-26 | 2023-04-25 | Tesla, Inc. | Optimizing neural network structures for embedded systems |
| US11562231B2 (en) | 2018-09-03 | 2023-01-24 | Tesla, Inc. | Neural networks for embedded devices |
| AU2019357615B2 (en) | 2018-10-11 | 2023-09-14 | Tesla, Inc. | Systems and methods for training machine models with augmented data |
| US11196678B2 (en) | 2018-10-25 | 2021-12-07 | Tesla, Inc. | QOS manager for system on a chip communications |
| US11816585B2 (en) | 2018-12-03 | 2023-11-14 | Tesla, Inc. | Machine learning models operating at different frequencies for autonomous vehicles |
| US11537811B2 (en) | 2018-12-04 | 2022-12-27 | Tesla, Inc. | Enhanced object detection for autonomous vehicles based on field view |
| US11610117B2 (en) | 2018-12-27 | 2023-03-21 | Tesla, Inc. | System and method for adapting a neural network model on a hardware platform |
| US11150664B2 (en) | 2019-02-01 | 2021-10-19 | Tesla, Inc. | Predicting three-dimensional features for autonomous driving |
| US10997461B2 (en) | 2019-02-01 | 2021-05-04 | Tesla, Inc. | Generating ground truth for machine learning from time series elements |
| US11567514B2 (en) | 2019-02-11 | 2023-01-31 | Tesla, Inc. | Autonomous and user controlled vehicle summon to a target |
| US10956755B2 (en) | 2019-02-19 | 2021-03-23 | Tesla, Inc. | Estimating object properties using visual image data |
| CN112651920A (en) * | 2020-07-24 | 2021-04-13 | 深圳市唯特视科技有限公司 | PCB bare board line flaw detection method and device and electronic equipment |
Citations (1)
| 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 |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10192301B1 (en) * | 2017-08-16 | 2019-01-29 | Siemens Energy, Inc. | Method and system for detecting line defects on surface of object |
| CN108021938A (en) * | 2017-11-29 | 2018-05-11 | 中冶南方工程技术有限公司 | A kind of Cold-strip Steel Surface defect online detection method and detecting system |
| CN108445011B (en) * | 2018-03-12 | 2021-06-11 | 苏州天准科技股份有限公司 | Defect detection system and method based on deep learning |
| CN108805102A (en) * | 2018-06-28 | 2018-11-13 | 中译语通科技股份有限公司 | A kind of video caption detection and recognition methods and system based on deep learning |
-
2019
- 2019-03-15 CN CN201910198350.6A patent/CN109975308B/en active Active
Patent Citations (1)
| 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 |
Non-Patent Citations (2)
| Title |
|---|
| 《Segmentation of welding defects using a robust algorithm》;Carrasco M A等人;《Materials Evaluation》;20041231;第62卷(第11期);第1142-1147页 * |
| 《基于深度学习的快销品图像识别系统设计与开发》;戴同武;《中国优秀硕士学位论文全文数据库 信息科技辑》;20190115(第01期);第11-36页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109975308A (en) | 2019-07-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109975308B (en) | Surface detection method based on deep learning | |
| CN111179225B (en) | Test paper surface texture defect detection method based on gray gradient clustering | |
| CN106780486B (en) | Steel plate surface defect image extraction method | |
| CN104574353B (en) | The surface defect decision method of view-based access control model conspicuousness | |
| MXPA06003027A (en) | Apparatus and method for automated web inspection. | |
| Islam et al. | Capsule image segmentation in pharmaceutical applications using edge-based techniques | |
| CN110827235B (en) | Steel plate surface defect detection method | |
| CN114693610A (en) | Welding seam surface defect detection method, equipment and medium based on machine vision | |
| CN110070523B (en) | Foreign matter detection method for bottle bottom | |
| CN113298769B (en) | FPC flexible flat cable appearance defect detection method, system and medium | |
| CN106780464A (en) | A kind of fabric defect detection method based on improvement Threshold segmentation | |
| CN106780437B (en) | A kind of quick QFN chip plastic packaging image obtains and amplification method | |
| CN111681213A (en) | Light guide plate line scratch defect detection method based on deep learning | |
| CN116468726B (en) | Online foreign matter line detection method and system | |
| CN113155839A (en) | Steel plate outer surface defect online detection method based on machine vision | |
| CN107610119A (en) | The accurate detection method of steel strip surface defect decomposed based on histogram | |
| CN115908415A (en) | Defect detection method, device and equipment based on edge and storage medium | |
| KR101929669B1 (en) | The method and apparatus for analyzing an image using an entropy | |
| CN110648330A (en) | Defect detection method for camera glass | |
| CN116228651A (en) | Cloth defect detection method, system, equipment and medium | |
| CN115953373A (en) | Glass defect detection method and device, electronic equipment and storage medium | |
| CN116843687A (en) | Communication optical cable surface flaw detection method and device | |
| Peng et al. | A fast detection scheme for original fabric based on Blob, canny and rotating integral algorithm | |
| CN114166849B (en) | Method for detecting defects of printed carbon lines and moisture-sensitive film of humidity sensor | |
| Balasubramani et al. | Defect identification in casting surface using image processing techniques |
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 | ||
| CP02 | Change in the address of a patent holder |
Address after: Room 3104, Xuanye building, Pioneer Park, torch hi tech Zone, Xiamen, Fujian, 361000 Patentee after: WEIKU (XIAMEN) INFORMATION TECHNOLOGY Co.,Ltd. Address before: 361000 unit A-8, unit 03, building D, Xiamen international shipping center, No. 97, Xiangyu Road, Xiamen area, China (Fujian) pilot Free Trade Zone, Xiamen, Fujian Patentee before: WEIKU (XIAMEN) INFORMATION TECHNOLOGY Co.,Ltd. |
|
| CP02 | Change in the address of a patent holder |