CN111738963A - Tobacco bale package appearance detection method based on deep self-learning - Google Patents
Tobacco bale package appearance detection method based on deep self-learning Download PDFInfo
- Publication number
- CN111738963A CN111738963A CN202010727482.6A CN202010727482A CN111738963A CN 111738963 A CN111738963 A CN 111738963A CN 202010727482 A CN202010727482 A CN 202010727482A CN 111738963 A CN111738963 A CN 111738963A
- Authority
- CN
- China
- Prior art keywords
- image
- images
- detection
- standard
- cigarette packet
- 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.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 76
- 235000002637 Nicotiana tabacum Nutrition 0.000 title claims abstract description 20
- 244000061176 Nicotiana tabacum Species 0.000 title 1
- 230000007547 defect Effects 0.000 claims abstract description 29
- 241000208125 Nicotiana Species 0.000 claims abstract description 19
- 238000005286 illumination Methods 0.000 claims abstract description 11
- 238000012937 correction Methods 0.000 claims abstract description 8
- 235000019504 cigarettes Nutrition 0.000 claims description 30
- 238000004364 calculation method Methods 0.000 claims description 13
- 230000008859 change Effects 0.000 claims description 9
- 238000001914 filtration Methods 0.000 claims description 9
- 238000000034 method Methods 0.000 claims description 7
- 230000011218 segmentation Effects 0.000 claims description 6
- 230000004927 fusion Effects 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 4
- 230000005540 biological transmission Effects 0.000 claims description 3
- 239000000428 dust Substances 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000013508 migration Methods 0.000 claims description 3
- 230000005012 migration Effects 0.000 claims description 3
- 238000007500 overflow downdraw method Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 230000009466 transformation Effects 0.000 claims description 3
- 238000007499 fusion processing Methods 0.000 claims description 2
- 238000012856 packing Methods 0.000 claims 1
- 238000004806 packaging method and process Methods 0.000 description 3
- 239000000123 paper Substances 0.000 description 3
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000002679 ablation Methods 0.000 description 1
- XAGFODPZIPBFFR-UHFFFAOYSA-N aluminium Chemical compound [Al] XAGFODPZIPBFFR-UHFFFAOYSA-N 0.000 description 1
- 229910052782 aluminium Inorganic materials 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000011888 foil Substances 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 239000005022 packaging material Substances 0.000 description 1
- 239000011087 paperboard Substances 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- 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
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration by the use of histogram techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration by the use of more than one image, e.g. averaging, subtraction
-
- G06T5/70—
-
- 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
- G06T7/97—Determining parameters from multiple pictures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/50—Extraction of image or video features by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
- G06V10/507—Summing image-intensity values; Histogram projection analysis
Abstract
The invention relates to a tobacco bale package appearance detection method based on deep self-learning, and belongs to the technical field of tobacco bale intelligent detection. Firstly, acquiring a standard image and generating a template based on an integral area NCC algorithm; and acquiring all defect images for training, and setting an allowable defect tolerance. And acquiring a detection image by using a double-station CCD sensor, and aligning, fusing and carrying out illumination transfer correction on the image. And finally, comparing the image with the template, and judging whether the detected image is qualified. The application overcomes the detection difficulty caused by light reflection, and has the advantages of wide detection range, simple operation, high detection precision and high detection accuracy.
Description
Technical Field
The invention relates to a tobacco bale package appearance detection method based on deep self-learning, and belongs to the technical field of tobacco bale intelligent detection.
Background
With the continuous progress of science and technology, the tobacco industry has higher and higher requirements on the cigarette packet packaging quality detection, and new challenges are provided for detection indexes, detection speed, six-surface full detection, reliability and the like. A lot of methods for detecting the package appearance based on the imaging principle exist at home and abroad, but the detection methods have the problems of low detection speed, incapability of detecting a reflective packaging material, complex arrangement, inconvenience in adjustment, instability, low detection accuracy and the like. Aiming at the defects in the prior art, the invention provides the cigarette packet package appearance detection method based on the depth self-learning, which is simple to use, high in detection speed and reliable in detection result.
Disclosure of Invention
In order to solve the technical problem, the invention provides a tobacco bale packaging appearance detection method based on depth self-learning, which comprises the following steps
Collecting images of six surfaces of at least three standard package appearances, adopting a generated template image based on an integral image, then calculating other standard images by using an NCC algorithm, and finally calculating a standard deviation according to a plurality of NCC calculation results;
collecting NG images of six faces of at least 20 unqualified package appearances, detecting the NG images by using an NCC algorithm, then finding out a defect area according to a standard deviation, obtaining geometric characteristic parameters of the defect area, and finally automatically setting defect tolerance according to the geometric characteristic parameters;
acquiring six-surface images of the cigarette packet to be detected by using a double-station CCD sensor, carrying out alignment, fusion and illumination migration correction on the images, carrying out NCC calculation on the obtained images, then finding out a defect area according to a standard deviation, and obtaining defect characteristic data of the cigarette packet to be detected;
and step four, comparing the defect characteristic data of the cigarette packet to be detected obtained in the step three with the template, if the defect characteristic data is within the allowable defect tolerance set in the step two, judging that the appearance of the cigarette packet to be detected is qualified, otherwise, judging that the appearance of the cigarette packet to be detected is unqualified, and sending an instruction of eliminating unqualified pieces.
Further, the NCC algorithm based on integral image in step 1 includes the steps of:
the standard template integral image formula of the sum calculated from the top to the bottom of the standard image and from left to right is as follows:
wherein ii (x, y) is expressed as the sum of gray values of all pixel points from (0, 0) to (x, y), i.e. an integral graph. i (x, y) represents the gray value of the pixel point at (x, y);
the formula for the calculation of NCC can be expressed as follows:
whereinRepresents the window size, anF denotes a template image, r denotes a detection image,representing the sum of squares of the template integral plots,representing the sum of squares of the suspect integrals,represents the mean of an arbitrary window of the template integral map,representing the mean of an arbitrary window of the integrated image to be examined.
And (4) calculating by using an NCC formula, and outputting a calculation result after five integral map indexes of the sum and the square sum of the image to be detected below the window and the template and the cross product of the sum and the square sum and the cross product of the image to be detected and the template are established by the integral image.
Further, in the step 3, only the standard image of at least three cigarette packets is needed, at least 20 NG cigarette packet images are needed when the automatic defect tolerance setting is started, and the manual modification can be performed when the automatic defect tolerance setting is not started.
Further, the alignment process in step three includes
Step a: tobacco bale gradient images obtained by using a cable filtering kernel;
step b: sorting the tobacco bale gradient strength to obtain five regions with the largest region gradient mean value;
step c: taking the gradient histograms of the images of the five regions as a matching feature descriptor HOG, searching a shift position which enables the HOG feature descriptor difference value of the standard image and the HOG feature descriptor difference value of the detection image to be minimum by a sliding window algorithm in the field of the deviation from the central range, and obtaining feature mapping point pair sets of 5 strong gradient regions;
step d: and finally, solving a change matrix by using a RANSAC algorithm to complete image alignment transformation.
Further, the fusion processing in the third step includes using a pyramid algorithm, obtaining highlight reflection areas of the two tobacco bale images collected by the double-station CCD sensor according to a highlight density distribution kernel function, and performing area replacement and fusion of highlight transmission areas of the two images by using a fusion method based on weighted average.
Further, the correction processing in step three includes
Step 1) carrying out fast discrete Fourier transform on the standard image to obtain a frequency domain image of the standard image, carrying out low-pass filtering on the frequency domain image, and carrying out inverse Fourier transform to obtain a light intensity distribution image of the image;
step 2) segmenting the light intensity distribution image by adopting a KNN algorithm, and calculating the light intensity mean value of each segmentation area of the standard image;
step 3) carrying out fast discrete Fourier transform on the detection image to obtain a frequency domain image of the detection image, carrying out low-pass filtering on the frequency domain image, and carrying out inverse Fourier transform to obtain a light intensity distribution image of the detection image;
step 4) segmenting the light intensity distribution image in the step 3) by adopting a KNN algorithm, and calculating the light intensity mean value of each segmentation area of the detection image;
step 5) obtaining the light intensity mean deviation ratio of the area standard image and the detection image;
and 6) carrying out illumination correction on the detected image according to the mean deviation ratio in the step 5), and reducing false detection caused by illumination difference.
Further, when the cigarette packet is detected on line in the third step, the change of light and dust of the system is detected, and if the change is qualified, the fourth step is executed; if not, judging whether the alarm signal reaches an alarm threshold value; if so, performing fault alarm; and if not, calibrating and returning to the first step.
The invention has the beneficial effects that: according to the invention, the detection images are acquired through binocular vision sensing, and the cigarette packet images based on the IGO-HOG images are quickly aligned to reduce errors; and then, image ablation and combination are carried out on the images, reflection is overcome, illumination function migration of illumination change of the cigarette packet is corrected, and accurate detection is carried out based on an NCC algorithm of an integral image. The invention has the advantages of wide detection range, simple operation, high detection precision and high detection accuracy.
Drawings
FIG. 1 is a schematic diagram of the logical structure of the present invention.
Detailed Description
The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic views illustrating only the basic structure of the present invention in a schematic manner, and thus show only the constitution related to the present invention.
As shown in fig. 1, the logical structure of the present invention is schematically illustrated.
Firstly, collecting images of six surfaces of at least three standard package appearances, adopting a generated template image based on an integral image, then calculating other standard images by using an NCC algorithm, and finally calculating a standard deviation according to a plurality of NCC calculation results;
collecting all the corresponding defect images of the six surfaces, detecting the NG images by using an NCC algorithm, then finding out a defect area according to the standard deviation, obtaining the geometric characteristic parameters of the defect area, and finally automatically setting the defect tolerance according to the geometric characteristic parameters.
The detection method comprises the following specific steps:
step one, storing appearance image characteristics of standard packages: and storing the acquired standard image and the standard deviation thereof in a computing device.
And (II) performing on-line comprehensive detection on the tobacco bale.
Firstly, two pairs of cigarette packet images of the same cigarette packet at different visual angles are obtained by using a double-station CCD sensor, and then the two pairs of cigarette packet images are aligned:
step a: tobacco bale gradient images obtained by using a cable filtering kernel;
step b: sorting the tobacco bale gradient strength to obtain five regions with the largest region gradient mean value;
step c: taking the gradient histograms of the images of the five regions as a matching feature descriptor HOG, searching a shift position which enables the HOG feature descriptor difference value of the standard image and the HOG feature descriptor difference value of the detection image to be minimum by a sliding window algorithm in the field of the deviation from the central range, and obtaining feature mapping point pair sets of 5 strong gradient regions;
step d: and finally, solving a change matrix by using a RANSAC algorithm to complete image alignment transformation.
Secondly, fusing two pairs of cigarette packet images which are aligned: based on a pyramid algorithm, obtaining highlight reflection areas of two tobacco bale images collected by a double-station CCD sensor according to a highlight density distribution kernel function, and carrying out area replacement and fusion of the highlight transmission areas of the two images based on a weighted average fusion method algorithm.
And finally, correcting the fused image:
step 1) carrying out fast discrete Fourier transform on the standard image to obtain a frequency domain image of the standard image, carrying out low-pass filtering on the frequency domain image, and carrying out inverse Fourier transform to obtain a light intensity distribution image of the image;
step 2) segmenting the light intensity distribution image by adopting a KNN algorithm, and calculating the light intensity mean value of each segmentation area of the standard image;
step 3) carrying out fast discrete Fourier transform on the detection image to obtain a frequency domain image of the detection image, carrying out low-pass filtering on the frequency domain image, and carrying out inverse Fourier transform to obtain a light intensity distribution image of the detection image;
step 4) segmenting the light intensity distribution image in the step 3) by adopting a KNN algorithm, and calculating the light intensity mean value of each segmentation area of the detection image;
step 5) obtaining the light intensity mean deviation ratio of the area standard image and the detection image;
and 6) carrying out illumination correction on the detected image according to the mean deviation ratio in the step 5), and reducing false detection caused by illumination difference.
Detecting system parameters: detecting no problem according to the change of the light and the dust of the detection system, and entering the step (IV); detecting the problem, judging whether the problem reaches an alarm threshold value, and if the problem reaches the alarm threshold value, performing alarm maintenance; if not, the standard parameters are automatically corrected, and the result judgment is carried out according to the corrected standard parameters in the detection result judgment process. In general, when a light source fault, an optical fiber fault, excessive rejection, detection interruption and camera abnormity occur, the processor sends out an alarm signal and controls the alarm to give an alarm.
And (IV) analyzing the detection result. And C, comparing the images obtained in the third step with the standard template.
Through the calculation of the formula, after the integral image is established, the sum and the square sum of the image to be detected below a window and the template and the cross product of the sum and the square sum are multiplied by five integral image indexes, the whole processing process is completed, the calculation result is finally output, and the NCC calculation result is compensated by standard deviation.
And (V) judging whether the detection result is qualified. Combining the allowable tolerance of the defects, and if the calculation result is smaller than the tolerance, judging that the product is qualified; otherwise, judging that the product is unqualified and rejecting the product.
When the method is used for testing, the defects capable of being detected comprise the following conditions.
1. The detection area of the shell breakage, falling, deformation and tearing is more than 2 × 2mm2And then the detection can be carried out.
2. The small ear is opened and inclined for 2mm2And then the detection can be carried out.
3. Brand pattern missing, unclear, off-set, color shift 2mm2And then the detection can be carried out.
4. The exposed detection area of the paperboard and the aluminum foil paper is more than 2 × 2mm2。
5. Dislocation of package is 2mm2And then the detection can be carried out.
6. The deformation detection area of the end face of the small bag is more than 2 × 2mm2。
7. The detection area of the foreign matters and the stains is more than 2 × 2mm2。
8. Facing slip reverse angle and falling reverse angle 2mm2And then the detection can be carried out.
9. The wrong card number is mixed with other cards.
10. And no packaging paper is used.
11. The wrapping paper is reversed.
12. The tobacco bale is inverted.
Therefore, the method has the advantages of wide detection range, simple operation, high detection precision and high detection accuracy.
It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
The meaning of "and/or" as used herein is intended to include both the individual components or both.
The term "connected" as used herein may mean either a direct connection between components or an indirect connection between components via other components.
In light of the foregoing description of the preferred embodiment of the present invention, many modifications and variations will be apparent to those skilled in the art without departing from the spirit and scope of the invention. The technical scope of the present invention is not limited to the content of the specification, and must be determined according to the scope of the claims.
Claims (7)
1. A tobacco bale packing appearance detection method based on depth self-learning is characterized in that: comprises that
Collecting images of six surfaces of at least three standard package appearances, adopting a generated template image based on an integral image, then calculating other standard images by using an NCC algorithm, and finally calculating a standard deviation according to a plurality of NCC calculation results;
collecting NG images of six faces of at least 20 unqualified package appearances, detecting the NG images by using an NCC algorithm, then finding out a defect area according to a standard deviation, obtaining geometric characteristic parameters of the defect area, and finally automatically setting an allowable defect tolerance according to the geometric characteristic parameters;
acquiring six-surface images of the cigarette packet to be detected by using a double-station CCD sensor, carrying out alignment, fusion and illumination migration correction on the images, carrying out NCC calculation on the obtained images, then finding out a defect area according to a standard deviation, and obtaining defect characteristic data of the cigarette packet to be detected;
and step four, comparing the defect characteristic data of the cigarette packet to be detected obtained in the step three with the template, if the defect characteristic data is within the allowable defect tolerance set in the step two, judging that the appearance of the cigarette packet to be detected is qualified, otherwise, judging that the appearance of the cigarette packet to be detected is unqualified, and sending an instruction of eliminating unqualified pieces.
2. The cigarette packet package appearance detection method based on depth self-learning of claim 1, characterized in that: in the first step, the standard template integral image formula obtained by calculating the sum from top to bottom and from left to right of the standard image is as follows:
wherein ii (x, y) represents the sum of the gray values of all the pixel points from (0, 0) to (x, y), i.e. the integral graph, and i (x, y) represents the gray value of the pixel point at (x, y);
the formula for the calculation of NCC can be expressed as follows:
whereinRepresents the window size, anF denotes a template image, r denotes a detection image,representing the sum of squares of the template integral plots,representing the sum of squares of the suspect integrals,represents the mean of an arbitrary window of the template integral map,arbitrary window for representing an integrated image to be examinedThe average value of (a) of (b),
and (4) calculating by using an NCC formula, and outputting a calculation result after five integral map indexes of the sum and the square sum of the image to be detected below the window and the template and the cross product of the sum and the square sum and the cross product of the image to be detected and the template are established by the integral image.
3. The cigarette packet package appearance detection method based on depth self-learning of claim 1, characterized in that: in the second step, standard images of at least three cigarette packets are needed, at least 20 NG cigarette packet images are needed when the automatic defect tolerance setting is started, and the standard images can be modified manually when the automatic defect tolerance setting is not started.
4. The cigarette packet package appearance detection method based on depth self-learning of claim 1, characterized in that: the alignment process in the third step comprises
Step a: tobacco bale gradient images obtained by using a cable filtering kernel;
step b: sorting the tobacco bale gradient strength to obtain five regions with the largest region gradient mean value;
step c: taking the gradient histograms of the images of the five regions as a matching feature descriptor HOG, searching a shift position which enables the HOG feature descriptor difference value of the standard image and the HOG feature descriptor difference value of the detection image to be minimum by a sliding window algorithm in the field of the deviation from the central range, and obtaining feature mapping point pair sets of 5 strong gradient regions;
step d: and finally, solving a change matrix by using a RANSAC algorithm to complete image alignment transformation.
5. The cigarette packet package appearance detection method based on depth self-learning of claim 1, characterized in that: and the fusion processing in the third step comprises the steps of using a pyramid algorithm, obtaining highlight reflection areas of two tobacco bale images collected by the double-station CCD sensor according to a highlight density distribution kernel function, and carrying out area replacement and fusion of two image highlight transmission areas by a fusion method based on weighted average.
6. The cigarette packet package appearance detection method based on depth self-learning of claim 1, characterized in that: the correction processing in the third step comprises
Step 1) carrying out fast discrete Fourier transform on the standard image to obtain a frequency domain image of the standard image, carrying out low-pass filtering on the frequency domain image, and carrying out inverse Fourier transform to obtain a light intensity distribution image of the image;
step 2) segmenting the light intensity distribution image by adopting a KNN algorithm, and calculating the light intensity mean value of each segmentation area of the standard image;
step 3) carrying out fast discrete Fourier transform on the detection image to obtain a frequency domain image of the detection image, carrying out low-pass filtering on the frequency domain image, and carrying out inverse Fourier transform to obtain a light intensity distribution image of the detection image;
step 4) segmenting the light intensity distribution image in the step 3) by adopting a KNN algorithm, and calculating the light intensity mean value of each segmentation area of the detection image;
step 5) obtaining the light intensity mean deviation ratio of the area standard image and the detection image;
and 6) carrying out illumination correction on the detected image according to the mean deviation ratio in the step 5), and reducing false detection caused by illumination difference.
7. The cigarette packet package appearance detection method based on depth self-learning of claim 1, characterized in that: when the cigarette packet is detected on line in the third step, the change of light and dust of the system is detected, and if the change is qualified, the fourth step is executed; if not, judging whether the alarm signal reaches an alarm threshold value; if so, performing fault alarm; and if not, calibrating and returning to the first step.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010727482.6A CN111738963B (en) | 2020-07-27 | 2020-07-27 | Tobacco bale package appearance detection method based on deep self-learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010727482.6A CN111738963B (en) | 2020-07-27 | 2020-07-27 | Tobacco bale package appearance detection method based on deep self-learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111738963A true CN111738963A (en) | 2020-10-02 |
CN111738963B CN111738963B (en) | 2021-02-23 |
Family
ID=72657741
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010727482.6A Active CN111738963B (en) | 2020-07-27 | 2020-07-27 | Tobacco bale package appearance detection method based on deep self-learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111738963B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114140400A (en) * | 2021-11-16 | 2022-03-04 | 湖北中烟工业有限责任公司 | Method for detecting cigarette packet seal defects based on RANSAC and CNN algorithms |
CN114348368A (en) * | 2022-01-29 | 2022-04-15 | 红云红河烟草(集团)有限责任公司 | Method and system for evaluating soft package cigarette package forming quality |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2019228868A1 (en) * | 2018-05-28 | 2019-12-05 | Gima S.P.A. | Surface quality inspection method and apparatus |
CN110987949A (en) * | 2019-12-24 | 2020-04-10 | 河南中烟工业有限责任公司 | Cigarette packet package appearance detection method |
CN111402238A (en) * | 2020-03-17 | 2020-07-10 | 苏州光领电子科技有限公司 | Defect identification system realized through machine vision |
-
2020
- 2020-07-27 CN CN202010727482.6A patent/CN111738963B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2019228868A1 (en) * | 2018-05-28 | 2019-12-05 | Gima S.P.A. | Surface quality inspection method and apparatus |
CN110987949A (en) * | 2019-12-24 | 2020-04-10 | 河南中烟工业有限责任公司 | Cigarette packet package appearance detection method |
CN111402238A (en) * | 2020-03-17 | 2020-07-10 | 苏州光领电子科技有限公司 | Defect identification system realized through machine vision |
Non-Patent Citations (1)
Title |
---|
GLOOMYFISH: "图像处理之积分图应用三(基于NCC快速相似度匹配算法", 《CSDN博客HTTPS://BLOG.CSDN.NET/JIA20003/ARTICLE/DETAILS/53021614》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114140400A (en) * | 2021-11-16 | 2022-03-04 | 湖北中烟工业有限责任公司 | Method for detecting cigarette packet seal defects based on RANSAC and CNN algorithms |
CN114140400B (en) * | 2021-11-16 | 2024-04-12 | 湖北中烟工业有限责任公司 | Method for detecting cigarette packet label defect based on RANSAC and CNN algorithm |
CN114348368A (en) * | 2022-01-29 | 2022-04-15 | 红云红河烟草(集团)有限责任公司 | Method and system for evaluating soft package cigarette package forming quality |
CN114348368B (en) * | 2022-01-29 | 2023-08-15 | 红云红河烟草(集团)有限责任公司 | Method and system for evaluating soft package cigarette package forming quality |
Also Published As
Publication number | Publication date |
---|---|
CN111738963B (en) | 2021-02-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
KR101298957B1 (en) | Wood knot detecting method, device, and program | |
CN109454006A (en) | Chemical fiber wire ingot is stumbled the device and its detection stage division of a defect on-line checking and classification | |
JP3041090B2 (en) | Appearance inspection device | |
CN111738963B (en) | Tobacco bale package appearance detection method based on deep self-learning | |
KR102168724B1 (en) | Method And Apparatus for Discriminating Normal and Abnormal by using Vision Inspection | |
KR20180118513A (en) | Sheet inspection device and inspection system | |
JP2009293999A (en) | Wood defect detector | |
CN111415339B (en) | Image defect detection method for complex texture industrial product | |
CN109462999B (en) | Visual inspection method based on learning through data balance and visual inspection device using same | |
CN112200790B (en) | Cloth defect detection method, device and medium | |
CN106501272A (en) | Machine vision scolding tin position detecting system | |
JP2018179698A (en) | Sheet inspection device | |
CN106546600A (en) | A kind of PET bottle multi-orientation detection method and apparatus based on machine vision | |
Fu et al. | Medicine glass bottle defect detection based on machine vision | |
CN114820439A (en) | PCB bare board defect detection system and method based on AOI | |
JPH08189904A (en) | Surface defect detector | |
CN116091506B (en) | Machine vision defect quality inspection method based on YOLOV5 | |
CN107643611A (en) | A kind of automatic checkout system of liquid crystal display and backlight assembly precision | |
CN116297199B (en) | Detection system for automatically detecting lithium battery parameters based on machine vision | |
WO2020119246A1 (en) | Highlight object surface defect detection system and method based on reflected light | |
Yu et al. | An anomaly detection system for transparent objects using polarized-image fusion technique | |
CN209606340U (en) | Transparent medicine bottle bottom buildup defect detecting system | |
CN109040724B (en) | Light spot distortion detection method and device of structured light projector and readable storage medium | |
KR20130031331A (en) | Glass bottle inspection device | |
CN106872483A (en) | Optical detection apparatus are solved because of the method for the aeration detection in transparent material |
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 |