CN114332091B - Printed matter defect detection method based on artificial intelligence - Google Patents
Printed matter defect detection method based on artificial intelligence Download PDFInfo
- Publication number
- CN114332091B CN114332091B CN202210256462.4A CN202210256462A CN114332091B CN 114332091 B CN114332091 B CN 114332091B CN 202210256462 A CN202210256462 A CN 202210256462A CN 114332091 B CN114332091 B CN 114332091B
- Authority
- CN
- China
- Prior art keywords
- texture
- deformation
- line
- printed matter
- image
- 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
Landscapes
- Image Processing (AREA)
Abstract
The invention relates to a presswork defect detection method based on artificial intelligence, which comprises the steps of obtaining a paper presswork image and a bottom plate image, carrying out gray level processing to obtain a gray level image, removing perspective deformation in the gray level image, obtaining key points which are matched with each other between two adjacent lines in the presswork gray level image and the presswork bottom plate gray level image, obtaining the deformation quantity of each line of texture according to the key points, constructing a deformation law function by utilizing the maximum category deformation quantity mean value, the deformation quantity equation and the deformation quantity parameter weight of each line of texture deformation quantity in the gray level image of the printed matter, calculating the deformation error value of each line of texture according to the deformation law function, performing deformation correction on the texture of each line in the presswork image by using the deformation error value of each line of texture, DNN defect detection is carried out on the corrected image of the printed matter, the method is intelligent and accurate, and the precision of the defect detection of the printed matter is improved.
Description
Technical Field
The application relates to the field of image transmission, in particular to a presswork defect detection method based on artificial intelligence.
Background
When the defect detection of the printed matter is carried out, the paper is generally driven to move by the rotation of the two parallel rollers, a snapshot camera is placed right above the paper to collect printed pictures on the paper, and whether the printed matter has printing defects or not is judged by matching the printing information in the collected pictures with the bottom plate information. However, the detection method has the defects that the paper is loosened and deformed due to the fluctuation or tension change of the roller equipment, the phenomenon can cause the printing textures of different paper areas to generate compression deformation with different degrees, the deformation can be mistakenly detected as the defects, the defect detection precision of the printed matter is reduced, the deformation influence caused by the speed fluctuation or the paper tension influence is not considered in the conventional defect detection of the printed matter, the defect detection is directly carried out, and the deformation caused by the loosening of the paper is mistakenly detected as the defects.
Disclosure of Invention
The invention provides a presswork defect detection method based on artificial intelligence, which solves the problem that the presswork defect detection does not consider speed fluctuation or false detection caused by paper tension, and adopts the following technical scheme:
respectively acquiring a gray level image of a paper printed matter and a gray level image of a printing matter bottom plate;
acquiring key points which are matched with each other between two adjacent lines in the gray-scale image of the printed matter and the gray-scale image of the printing master, and acquiring the deformation quantity of each line of texture in the gray-scale image of the printed matter by using the vector of the key points between two adjacent lines in the gray-scale image of the printed matter and the vector of the key points between two corresponding adjacent lines in the gray-scale image of the printing master;
obtaining a deformation quantity error value of each line of texture by using the deformation quantity of each line of texture in the gray level image of the printed matter;
and carrying out deformation correction on the texture of each line in the printed matter image by using the deformation error value of each line of texture, and carrying out defect detection on the corrected printed matter image.
The method for acquiring the deformation quantity of each line of the texture comprises the following steps:
acquiring corresponding key points in the grey-scale image of the printed matter and the grey-scale image of the master plate through corner matching;
dividing the printed matter gray level image and the master gray level image into a plurality of lines through equidistant parallel lines;
respectively combining each key point in every two adjacent lines in the grey-scale image of the printed matter and the grey-scale image of the master plate with the key point which is closest to the key point in the first line and the second line into a texture;
subtracting the coordinates of the key points in the first row from the coordinates of the key points in the second row in each texture to obtain the vector of the texture;
subtracting the vector of the texture in the gray image of the printed matter from the vector of the texture in the image of the master to obtain the deformation amount of the texture;
and taking every two adjacent lines as a texture line to obtain the deformation quantity of the texture of each line.
The method for obtaining the deformation error value of each line of texture comprises the following steps:
constructing a texture deformation quantity rule function:
in the formula (I), the compound is shown in the specification,
as a function of the deformation law of the texture, k =1,2, 3.. and M, M is the number of lines,
for the deformation equation of the texture of the k-th line,
the maximum class mean of the deformation quantities representing the texture of the k-th line,
the deformation parameter weight of the k-th line texture,
the function constraint condition is the deformation error value of the k-th line texture
Wherein, in the step (A),
is a unit transverse vector;
initially randomly setting a deformation error value of each line of texture in a gray scale image of a printed matter to obtain a deformation error value sequence;
fitting the deformation equation of each line of texture according to the deformation error value sequence to obtain the deformation fitted to the kth line of texture, namely
Function values;
in this case, the product contains only unknown quantity
The texture deformation law function of (1);
updating the sequence of the deformation error values by using a gradient descent method, repeating the steps to obtain a texture deformation law function,obtaining the optimal solution until the texture deformation quantity rule function is converged, and calculating
。
Maximum class mean of deformation amount of each line of texture
The acquisition method comprises the following steps:
carrying out density clustering on deformation of the k-th line of textures, then screening out a texture deformation set of the largest category and solving the expectation of the set, wherein the expectation is
。
The weight of the deformation parameter of each line of texture
The calculation method comprises the following steps:
in the formula (I), the compound is shown in the specification,
k =1,2,3,.. for the deformation parameter weight of the texture of the kth line, M is the number of lines in the gray scale map,
the distribution dispersion of the largest class distortion of the k-th line texture,
the dispersion of the deformation amount distribution for the t-th line texture,
calculation method and
and (5) the consistency is achieved.
The method for calculating the distribution dispersion of the maximum class deformation quantity of each line of texture comprises the following steps:
in the formula (I), the compound is shown in the specification,
the distribution dispersion of the largest class distortion of the k-th line texture,
for the deformation of the ith texture in the k line of textures,
represents the mean value of the deformation quantity of the maximum category after the density clustering is carried out on all the texture points in the kth row,
is the number of textures in the k-th line of textures
The method for obtaining the deformation equation of each line of texture comprises the following steps:
the deformation error value of each line of texture in the deformation error value sequence and the maximum class mean value of the deformation of each line of texture
Adding to obtain the standard deformation of each line of texture;
obtaining a standard deformation quantity sequence consisting of standard deformation quantities of each line of textures, and calculating the modular length of each vector element in the sequence to obtain a modular length sequence;
and fitting a shape variable equation by using a least square method by using the texture row sequence as an independent variable and the modular length sequence as a dependent variable.
The method for performing deformation correction on the texture of each line comprises the following steps:
and adding the deformation amount of each texture in each line of textures and the deformation amount error value of the texture corresponding to each line to realize deformation correction of each texture.
The method for detecting the defects of the corrected image of the printed matter comprises the following steps:
and positioning the printing defects in the corrected printed product image by utilizing a DNN mode.
The invention has the beneficial effects that: deformation interference caused by inconsistent relative motion of paper and a camera during detection is eliminated by analyzing the deformation rule, and then the defect detection precision of the printed matter is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a schematic flow chart of an artificial intelligence based method for detecting defects in printed matter according to the present invention;
FIG. 2 is a schematic diagram of a usage scenario of an artificial intelligence based presswork defect detection method of the present invention;
FIG. 3 is a schematic illustration of a deformed sheet of paper for an artificial intelligence based presswork defect detection method of the present invention;
FIG. 4 is a schematic diagram of a texture in the artificial intelligence-based method for detecting defects in a printed product according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
An embodiment of the method for detecting defects of printed products based on artificial intelligence of the present invention, as shown in fig. 1, includes:
the method comprises the following steps: respectively acquiring a gray level image of a paper printed matter and a gray level image of a printing master plate;
the purpose of this step is to process the print image to exclude the effects of perspective distortion.
The use scene of this embodiment is as shown in fig. 2, and the printing paper is dragged to move through two rollers of rotation, sets up a snapshot camera directly over the printing paper and gathers the printed matter image, realizes the printed matter defect detection through comparing the comparison of printed matter image and photographic plate information.
Firstly, set up the snap camera directly over the printed matter, rotate through two upper and lower parallel running rollers and drive the printed matter paper and remove in real time, the camera is through the intermittent type snap gather the printed matter image, the intermittent type time of camera can be according to gyro wheel slew velocity adjustment, the speed of gyro wheel should not be too fast, places and has the removal fuzzy phenomenon. Graying treatment: converting the captured print image from RGB color space to grayscale space to obtain grayscale image
。
Then, carrying out perspective deformation restoration on the gray-scale image:
in order to prevent the perspective deformation with large middle and small periphery in the image of the printed matter from influencing the speed fluctuation deformation at the back, the perspective deformation in the printed matter needs to be removed firstly, and a perspective deformation conversion model is constructed according to the relative position relationship between a camera and paper and is removed through the perspective model.
Finally, the picture obtained through the processing only contains the deformation caused by the speed fluctuation and the printing defect deformation, so that the accurate printing defect detection can be realized only by eliminating the speed fluctuation deformation, and the deformation caused by the speed fluctuation is shown in fig. 3.
Step two: acquiring key points which are matched with each other between two adjacent lines in the gray-scale image of the printed matter and the gray-scale image of the printing master, and acquiring the deformation quantity of each line of texture in the gray-scale image of the printed matter by using the vector of the key points between two adjacent lines in the gray-scale image of the printed matter and the vector of the key points between two corresponding adjacent lines in the gray-scale image of the printing master;
the purpose of this step is to calculate the amount of deformation of each texture by taking the texture in the image.
The method for acquiring the deformation quantity of each texture comprises the following steps:
(1) and carrying out corner matching on the collected gray-scale image of the printed matter and the gray-scale image of the bottom plate by matching with a harris corner matching algorithm to obtain a group of matching points of the gray-scale image of the printed matter and the gray-scale image of the bottom plate.
(2) Dividing the two images into lines: parallel lines are made at intervals of 10 pixels in parallel with the wide side of the print, and the gray-scale image of the print and the gray-scale image of the master are divided into a plurality of lines by the parallel lines.
(3) In a printed matter gray-scale image and a master gray-scale image respectively, forming a texture by each key point in a first row and the key point which is closest to the key point in a second row in every two adjacent rows, wherein as shown in fig. 4, 1 is a first row, 2 is a second row, a key point A and a key point B form a texture, and a key point C and a key point D form a texture;
(4) in the gray-scale image of the printed matter, subtracting the coordinate of the key point in the row two from the coordinate of the key point in the row one in each texture to obtain the vector of the texture, wherein as shown in fig. 4, the coordinate of the key point a subtracted from the coordinate of the key point B is the vector of the texture, and similarly, the vector of the corresponding texture is also obtained in the gray-scale image of the bottom plate;
(5) taking every two lines as a texture line, as shown in fig. 4, 1 and 2 form a texture line, and a vector of each texture in each line of textures can be obtained;
(6) subtracting the vector of each texture in the grey-scale image of the printed matter from the vector of the texture in the master image to obtain the deformation amount of the texture:
in the formula (I), the compound is shown in the specification,
for the deformation of the ith texture in the k line of textures,
the vector of the ith texture in the k line of textures in the gray-scale map of the print,
the vector of the ith texture in the k line texture in the master gray scale image.
(7) According to (6), the deformation amount of each line of texture, that is, the deformation amount of each texture in each line of texture, can be obtained.
Step three: obtaining a deformation quantity error value of each line of texture by using the deformation quantity of each line of texture in the gray level image of the printed matter;
the purpose of the step is to analyze the deformation quantity of the texture of the same line to construct a deformation law function of the deformation quantity of the texture of each line, and calculate the error value of the deformation quantity of the texture of each line according to the deformation law function of the texture of each line.
The method comprises the following steps of:
(1) constructing a texture deformation rule function:
the function constraint is:
in the formula (I), wherein,
k =1,2,3,.. for the deformation law function of the texture of the kth line, M is the line number,
for the deformation equation of the texture of the k-th line,
the maximum class mean of the deformation quantities representing the texture of the k-th line,
the deformation parameter weight of the k-th line texture,
the deformation error value of the k-th line of texture;
wherein, obtaining
The method comprises the following steps: carrying out density clustering on the type variables of the k-th line of textures, screening out the maximum (most) type deformation quantity set, and solving the expectation of the type deformation quantity set, wherein the expectation is the expectation
，
It should be noted that the reason for obtaining the value in this way is that, in this embodiment, it is assumed that the deformation amount of the texture of the entire line caused by the paper deformation is the same, and the print defects in the entire line are different, so it is considered that the distribution of the texture deformation caused by the paper deformation in the entire line is relatively concentrated, and therefore, the value of the texture deformation caused by the paper deformation is expected through the deformation of a relatively large category.
Wherein the content of the first and second substances,
the weight of the deformation parameter of the k-th line texture, and the deformation parameter weight is used to reflect
The lower the accuracy of the value, the smaller the weight, the more the accuracy of the value is, the more the amount of deformation of the texture due to deformation of the paper in the entire line should be as equal as possible, i.e. the higher the concentration of the deformation distribution, thus reflecting the accuracy of the line of deformation values by the concentration of the maximum class deformation distribution.
Herein, the
The calculation steps are as follows:
firstly, calculating the distribution dispersion of the maximum category deformation:
in the formula (I), the compound is shown in the specification,
the distribution dispersion of the deformation amount of the largest category after density clustering of the texture of the k-th row,
the amount of deformation of the ith texture in the kth line,
and representing the mean value of the deformation quantity of the maximum category after all the texture point densities in the k-th row are clustered.
Representing the number of textures in the k-th line of textures.
Then, according to the distribution dispersion of the maximum category deformation quantity, obtaining the deformation quantity parameter weight of the k-th line of texture:
in the formula (I), the compound is shown in the specification,
the deformation parameter weight of the k-th line texture,
the dispersion of the deformation amount distribution of the texture of the t-th row,
calculation method and
and (5) the consistency is achieved.
Wherein the content of the first and second substances,
is the error value (by deformation amount) of the texture deformation amount of the k-th line obtained by deformation distribution data
As the difference between the amount of texture deformation caused by deformation of the line of sheets and the standard amount of texture deformation caused by deformation of the line of sheets).
Wherein the content of the first and second substances,
for the deformation equation of each line texture, because the deformation form of the loose paper of the paper is similar to the scene of the catenary model, the basic mathematical model of the deformation equation is the catenary equation, and the independent variable of the deformation equation is the line number
Dependent variable is standard deformation
The fitting method of the deformation equation is a least square method.
Wherein the content of the first and second substances,
the deformation vector modulo length, which represents the texture of the k-th line, also represents the value of the standard texture deformation caused by the deformation of the paper in the k-th line,
the difference between the deformation variable value calculated by the deformation equation fitting at the k-th line and the standard deformation variable value is expressed due to the deformation equation fitting at the k-th lineThe deformation quantity variation law is described as accurately as possible, so that the value difference should be made as small as possible.
Wherein the content of the first and second substances,
the weighted error values representing the texture of line k should be made as small as possible since the function of the deformation quantity is to be made more accurate.
Wherein the content of the first and second substances,
the unit of the transverse vector is represented,
representing the projection of the deformation of the texture of the k-th line in the transverse direction, the deformation in the transverse direction should be 0, since the deformation of the paper only causes the longitudinal compression deformation of the printed texture.
(2) And calculating the deformation error value of each line of texture by using the deformation law function:
in this embodiment, a lagrangian function is constructed by a lagrangian multiplier method, and a deformation error value of each line of texture when a target function takes a minimum value is solved
The specific method comprises the following steps:
a. randomly giving an initial deformation error sequence
Each error value in the sequence is a deformation quantity error value of each line of texture;
b. obtaining a fitted warp value for each line of texture from the sequence of warp error values using the warp equation, i.e.
Function values;
c. in this case, the product contains only unknown quantity
The deformation law function of the texture;
d. updating the initial deformation error value sequence by using a gradient descent method, repeating the steps (1) to (3) to obtain the deformation law function of the kth line of textures until the function is converged to obtain an optimal solution, and calculating the error value of the texture deformation quantity of each line of textures
。
Wherein, the equation of deformation
The acquisition method comprises the following steps:
since the catenary equation is conventionally expressed as
The hyperparameter of the equation is
In the process of calculating the optimal solution of the deformation law function in the step, a deformation quantity error sequence is obtained;
(1) the deformation error value of each line of texture in the deformation error value sequence and the maximum class mean value of the deformation of each line of texture
Adding to obtain the standard deformation of each line of texture
；
(2) Obtaining a standard deformation quantity sequence consisting of standard deformation quantities of each line of textures, and calculating the modular length of each vector element in the sequence to obtain a modular length sequence;
(3) and fitting a deformation equation of the gray level image of the printed matter by using a least square method by using the texture row sequence as an independent variable and the die length sequence as a dependent variable.
Step four: and carrying out deformation correction on the texture of each line in the printed matter image by using the deformation error value of each line of texture, and carrying out defect detection on the corrected printed matter image.
The purpose of the step is to correct each line of texture by using the deformation error value of each line of texture obtained in the step three, and further detect and position defects, thereby improving the detection precision.
The method for correcting each line of texture comprises the following steps:
by comparing the deformation error value of each texture in each line of textures with that of the line of textures
And adding the textures to realize deformation correction of each texture.
The method for further detecting the defects comprises the following steps:
after the texture deformation caused by the paper deformation is removed, only printing defects exist in the printing image, so that the defect positioning can be realized by using a DNN (digital non-network) mode.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and should not be taken as limiting the scope of the present invention, which is intended to cover any modifications, equivalents, improvements, etc. within the spirit and scope of the present invention.
Claims (9)
1. A presswork defect detection method based on artificial intelligence is characterized by comprising the following steps:
respectively acquiring a gray level image of a paper printed matter and a gray level image of a printing matter bottom plate;
acquiring key points which are matched with each other between two adjacent lines in the gray-scale image of the printed matter and the gray-scale image of the printing master, and acquiring the deformation quantity of each line of texture in the gray-scale image of the printed matter by using the vector of the key points between two adjacent lines in the gray-scale image of the printed matter and the vector of the key points between two corresponding adjacent lines in the gray-scale image of the printing master;
obtaining a deformation quantity error value of each line of texture by using the deformation quantity of each line of texture in the gray level image of the printed matter;
and carrying out deformation correction on the texture of each line in the printed matter image by using the deformation error value of each line of texture, and carrying out defect detection on the corrected printed matter image.
2. The method for detecting the defect of the printed matter based on the artificial intelligence as claimed in claim 1, wherein the method for obtaining the deformation quantity of each line of the texture is as follows:
acquiring corresponding key points in the grey-scale image of the printed matter and the grey-scale image of the master plate through corner matching;
dividing the printed matter gray level image and the master gray level image into a plurality of lines through equidistant parallel lines;
respectively combining each key point in every two adjacent lines in the grey-scale image of the printed matter and the grey-scale image of the master plate with the key point which is closest to the key point in the first line and the second line into a texture;
subtracting the coordinates of the key points in the first row from the coordinates of the key points in the second row in each texture to obtain the vector of the texture;
subtracting the vector of the texture in the gray image of the printed matter from the vector of the texture in the image of the master to obtain the deformation amount of the texture;
and taking every two adjacent lines as a texture line to obtain the deformation quantity of the texture of each line.
3. The method for detecting the defects of the printed matters based on the artificial intelligence as claimed in claim 2, wherein the deformation error value of each line of the textures is obtained through a constructed texture deformation law function, and the expression of the texture deformation law function is as follows:
in the formula (I), the compound is shown in the specification,
k =1,2,3,. for a texture deformation law function, M is the number of lines,
for the deformation equation of the texture of the k-th line,
the maximum class mean of the deformation quantities representing the texture of the k-th line,
the deformation parameter weight of the k-th line texture,
the function constraint condition is the deformation error value of the k-th line texture
Wherein, in the step (A),
is a unit transverse vector;
initially randomly setting a deformation error value of each line of texture in a gray scale image of the printed matter to obtain a deformation error value sequence;
fitting the deformation equation of each line of texture according to the deformation error value sequence to obtain the deformation quantity fitted by the k-th line of texture, namely
Function values;
in this case, the product contains only unknown quantity
The texture deformation law function of (1);
updating the sequence of the deformation error values by using a gradient descent method, repeating the steps to obtain a texture deformation quantity regular function, obtaining an optimal solution when the texture deformation quantity regular function is converged, and calculating
。
4. The method as claimed in claim 3, wherein the maximum class mean of the deformation of each line of texture is
The acquisition method comprises the following steps:
carrying out density clustering on the deformation quantity of the k-th line texture, then screening out the texture deformation quantity set of the maximum category and solving the expectation of the set, wherein the expectation is
。
5. The method as claimed in claim 3, wherein the deformation parameter weight of each line of texture is determined by the artificial intelligence-based method for detecting defects in printed matter
The calculation method comprises the following steps:
in the formula (I), the compound is shown in the specification,
k =1,2,3,.. for the deformation parameter weight of the texture of the kth line, M is the number of lines in the gray scale map,
the distribution dispersion of the largest class distortion of the k-th line texture,
the dispersion of the deformation amount distribution for the t-th line texture,
calculation method and
and (5) the consistency is achieved.
6. The method for detecting the defect of the printed matter based on the artificial intelligence as claimed in claim 5, wherein the method for calculating the distribution dispersion of the maximum class deformation quantity of each line of the texture is as follows:
in the formula (I), the compound is shown in the specification,
the distribution dispersion of the largest class distortion of the k-th line texture,
for the deformation of the ith texture in the k line of textures,
represents the mean value of the deformation quantity of the largest category after all texture points in the k row are subjected to density clustering,
is the number of textures in the k-th line of textures.
7. The method for detecting the defect of the printed matter based on the artificial intelligence as claimed in claim 3, wherein the deformation equation of each line of the texture is obtained by:
the deformation error value of each line of texture in the deformation error value sequence and the maximum class mean value of the deformation of each line of texture
Adding to obtain the standard deformation of each line of texture;
acquiring a standard deformation sequence consisting of standard deformation of each line of texture, and calculating the modular length of each vector element in the sequence to obtain a modular length sequence;
and fitting a shape variable equation by using a least square method by using the texture row sequence as an independent variable and the modular length sequence as a dependent variable.
8. The method for detecting the defect of the printed matter based on the artificial intelligence as claimed in claim 1, wherein the method for performing deformation correction on the texture of each line comprises the following steps:
and adding the deformation amount of each texture in each line of textures and the deformation amount error value of the texture corresponding to each line to realize deformation correction of each texture.
9. The method for detecting the defect of the printed matter based on the artificial intelligence as claimed in claim 1, wherein the method for detecting the defect of the corrected printed matter image comprises the following steps:
and positioning the printing defects in the corrected printed product image by utilizing a DNN mode.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210256462.4A CN114332091B (en) | 2022-03-16 | 2022-03-16 | Printed matter defect detection method based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210256462.4A CN114332091B (en) | 2022-03-16 | 2022-03-16 | Printed matter defect detection method based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114332091A CN114332091A (en) | 2022-04-12 |
| CN114332091B true CN114332091B (en) | 2022-06-03 |
Family
ID=81033015
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210256462.4A Active CN114332091B (en) | 2022-03-16 | 2022-03-16 | Printed matter defect detection method based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114332091B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114913176B (en) * | 2022-07-18 | 2022-09-13 | 江苏启航箱包有限公司 | Flexible leather material scab defect detection method and system based on artificial intelligence |
| CN115311266B (en) * | 2022-10-10 | 2023-04-14 | 南通富兰妮纺织品有限公司 | Printing and dyeing identification method in textile printing and dyeing process |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5678595B2 (en) * | 2010-11-15 | 2015-03-04 | 株式会社リコー | INSPECTION DEVICE, INSPECTION METHOD, INSPECTION PROGRAM, AND RECORDING MEDIUM CONTAINING THE PROGRAM |
| JP6613214B2 (en) * | 2016-07-29 | 2019-11-27 | 富士フイルム株式会社 | Printed matter inspection apparatus, printed matter inspection method and program |
| CN108982525A (en) * | 2018-08-23 | 2018-12-11 | 云南省印刷技术研究所 | A kind of analysis and detection method for books and periodicals green fine printing quality |
| CN111127417B (en) * | 2019-12-20 | 2024-02-06 | 江苏理工学院 | Printing defect detection method based on SIFT feature matching and SSD algorithm improvement |
| CN111145163B (en) * | 2019-12-30 | 2021-04-02 | 深圳市中钞科信金融科技有限公司 | Paper wrinkle defect detection method and device |
| CN112132781A (en) * | 2020-08-19 | 2020-12-25 | 扬州哈工科创机器人研究院有限公司 | Texture defect detection method and device |
-
2022
- 2022-03-16 CN CN202210256462.4A patent/CN114332091B/en active Active
Non-Patent Citations (2)
| Title |
|---|
| 基于Siamese-YOLOv4的印刷品缺陷目标检测;楼豪杰;《计算机应用》;20210831(第11期);全文 * |
| 基于改进梯度幅值的包装缺陷检测算法研究及应用;宋丽梅;《应用光学》;20190715(第04期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114332091A (en) | 2022-04-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114332091B (en) | Printed matter defect detection method based on artificial intelligence | |
| CN107016647B (en) | Panoramic picture color tone consistency correcting method and system | |
| US20200133182A1 (en) | Defect classification in an image or printed output | |
| CN111862194B (en) | Deep learning plant growth model analysis method and system based on computer vision | |
| CN111402226A (en) | Surface defect detection method based on cascade convolution neural network | |
| CN107038416B (en) | Pedestrian detection method based on binary image improved HOG characteristics | |
| CN112818969A (en) | Knowledge distillation-based face pose estimation method and system | |
| CN114913138A (en) | Method and system for detecting defects of pad printing machine product based on artificial intelligence | |
| CN116228780A (en) | Silicon wafer defect detection method and system based on computer vision | |
| CN115564749A (en) | Method for constructing multi-class texture surface defect detection model based on lifelong learning | |
| CN115862131A (en) | Video data screening method for human motion recognition | |
| CN114266899A (en) | Image target parallel detection method based on multi-core DSP | |
| CN114612399A (en) | Picture identification system and method for mobile phone appearance mark | |
| JP3608106B2 (en) | Degradation evaluation system for steel surface using image processing | |
| CN110632094B (en) | Pattern quality detection method, device and system based on point-by-point comparison analysis | |
| CN115830514B (en) | Whole river reach surface flow velocity calculation method and system suitable for curved river channel | |
| CN115311266B (en) | Printing and dyeing identification method in textile printing and dyeing process | |
| CN116645351A (en) | Online defect detection method and system for complex scene | |
| CN114863258B (en) | Method for detecting small target based on visual angle conversion in sea-sky-line scene | |
| CN108182492B (en) | Data assimilation method and device | |
| CN113592870B (en) | Printing defect detection method based on self-adaptive focal length | |
| CN113610940B (en) | Ocean vector file and image channel threshold based coastal area color homogenizing method | |
| CN112911164A (en) | Exposure time adjusting method and system for camera for air tightness test | |
| CN108492289B (en) | Digital image quality evaluation system | |
| CN111260625A (en) | Automatic extraction method for offset printing large image detection area |
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 |