CN116309537A - Defect detection method for oil stain on surface of tab die - Google Patents
Defect detection method for oil stain on surface of tab die Download PDFInfo
- Publication number
- CN116309537A CN116309537A CN202310443245.0A CN202310443245A CN116309537A CN 116309537 A CN116309537 A CN 116309537A CN 202310443245 A CN202310443245 A CN 202310443245A CN 116309537 A CN116309537 A CN 116309537A
- Authority
- CN
- China
- Prior art keywords
- actual edge
- edge line
- point
- actual
- value
- 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
- 230000007547 defect Effects 0.000 title claims abstract description 39
- 238000001514 detection method Methods 0.000 title claims abstract description 38
- 238000000034 method Methods 0.000 claims abstract description 48
- 238000012216 screening Methods 0.000 claims abstract description 33
- 238000012545 processing Methods 0.000 claims abstract description 15
- 230000002159 abnormal effect Effects 0.000 claims description 60
- 238000010606 normalization Methods 0.000 claims description 26
- 238000013507 mapping Methods 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000003708 edge detection Methods 0.000 claims description 6
- 238000005457 optimization Methods 0.000 abstract description 6
- 230000000694 effects Effects 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000002068 genetic effect Effects 0.000 description 2
- 229910052751 metal Inorganic materials 0.000 description 2
- 239000002184 metal Substances 0.000 description 2
- HBBGRARXTFLTSG-UHFFFAOYSA-N Lithium ion Chemical compound [Li+] HBBGRARXTFLTSG-UHFFFAOYSA-N 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 239000004020 conductor Substances 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000007599 discharging Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 229910001416 lithium ion Inorganic materials 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 150000002739 metals Chemical class 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000010187 selection method Methods 0.000 description 1
- 239000004551 spreading oil Substances 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
- G06T5/90—Dynamic range modification of images or parts thereof
-
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/761—Proximity, similarity or dissimilarity measures
-
- 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
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
- Y02E60/10—Energy storage using batteries
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Quality & Reliability (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of image processing, in particular to a defect detection method for oil stains on the surface of a tab die. According to the method, an initial edge line is obtained through detection of a preset high threshold value of a canny operator, an actual edge line is obtained through gray level change difference and relative position screening of the initial edge line, an actual edge line group is obtained according to position distribution and gray level difference matching, a preset high threshold value is adjusted according to gradient change of the actual edge line in the actual edge line group to obtain an optimized high threshold value, an actual edge point is obtained according to the optimized high threshold value, the preset high threshold value is continuously adjusted to complete connection, an actual edge line to be detected of the actual edge group is obtained, and the actual edge line is subjected to iterative matching connection until a closed area is obtained, and an oil-stain area is screened according to gray level change rules. According to the invention, through image processing, the high threshold value is subjected to local self-adaptive optimization through gradient information and spatial information, so that a more complete and accurate oil stain region is obtained.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a defect detection method for oil stains on the surface of a tab die.
Background
The battery tab is an important component in a lithium ion battery, the battery is divided into a positive electrode and a negative electrode, and the tab is a metal conductor which leads the positive electrode and the negative electrode out of a battery core and is a key component for charging and discharging the battery. The battery electrode is connected to an external circuit through tabs, which are made of various metals, are thin and soft, and are very easily damaged. When defects appear in the mold for producing the tab, the quality of the tab can be greatly influenced, so that the tab mold needs to be detected for the surface defect condition, the influence of factors such as bulges or scratches is avoided, and the quality of the produced tab is unqualified.
When the surface defects of the tab mold are detected, the detected defects have lower precision because oil stains and bright spots existing in the surface can interfere with the identification of the defects. Generally, a canny operator is often used for detecting and extracting a defect area, in order to better remove the interference of oil stains, the oil stain area needs to be segmented, the contrast ratio of the oil stains to the background is low, and the oil stains are difficult to segment from the background completely. In the traditional algorithm, the canny operator threshold is often set manually and is also obtained through a genetic algorithm, but the diversity of the population is possibly degraded due to the degradation phenomenon of the genetic algorithm in the searching process, the convergence speed is slow, the finally obtained threshold is a local optimal solution rather than a global optimal solution, the searching time is long, and the efficiency is low. The optimization of the canny operator is limited to the self-adaptive optimization of the whole edge, and the adjustment of the local threshold value does not consider the change characteristics of different gray values, so that the obtained edge is inaccurate, and the local edge is a discontinuous edge section, so that the extraction of the region is not accurate and complete.
Disclosure of Invention
In order to solve the technical problems that the obtained edge is inaccurate and discontinuous in part and the extraction of the area is not accurate and complete in the prior art, the invention aims to provide a defect detection method for the greasy dirt on the surface of a tab die, and the adopted technical scheme is as follows:
the invention provides a defect detection method for oil stains on the surface of a tab die, which comprises the following steps:
obtaining a gray level image of the surface of a tab mold, and carrying out canny algorithm edge detection on the gray level image of the surface of the tab mold through a preset high threshold value to obtain an initial edge line; obtaining abnormal point probability of each pixel point according to gray value variation difference and relative distance between each pixel point and other pixel points on the initial edge line, and screening abnormal points on the initial edge line according to the abnormal point probability to obtain an actual edge line;
obtaining matching degree through gray value difference and position distribution between the actual edge lines, and screening out an actual edge line group formed by two similar actual edge lines according to the matching degree;
taking any actual edge line group as a detection group, and in an unconnected area between the actual edge lines in the detection group, starting from an endpoint of one actual edge line in the detection group, adjusting a preset high threshold according to gradient change of adjacent pixel points on the actual edge line to obtain an optimized high threshold of the endpoint; screening through the optimized high threshold value to obtain an actual edge point adjacent to the end point in the unconnected area, and obtaining a new actual edge line; starting from the actual edge points of the new actual edge lines, continuously adjusting a preset high threshold value to obtain all the actual edge points, and completing connection to obtain the actual edge lines to be detected of a detection group; obtaining the actual edge lines to be detected corresponding to each actual edge line group;
Iteratively matching and connecting all the actual edge lines to be detected until a closed area is obtained; and screening out the oil-polluted area according to the gray value change rule in the closed area.
Further, the obtaining of the outlier probability of each pixel point includes:
obtaining the maximum gray value difference between each pixel point on the initial edge line and the neighborhood pixel point in a preset neighborhood range;
taking one pixel point on the initial edge line as a reference pixel point, calculating and normalizing gray value differences between the reference pixel point and other pixel points on the initial edge line to obtain reference gray value differences;
taking the difference of the maximum gray value difference between the reference pixel point and other pixel points on the initial edge line as a gradient difference; multiplying the reference gray value difference by the gradient difference, and taking the product as a gray value change difference;
performing negative correlation mapping and normalization on Euclidean distances between the reference pixel point and other pixel points on the initial edge line to obtain an adjustment weight; and obtaining the gray value variation difference between the reference pixel point and all other pixel points, adjusting the gray value variation difference through the adjusting weight, and taking the average value of all the weighted gray value variation differences as the probability of the abnormal point.
Further, the obtaining of the matching degree includes:
acquiring the end points of each actual edge line, performing negative correlation mapping and normalization processing on the minimum value of Euclidean distance between the two actual edge line end points, and acquiring a distance distribution characteristic value;
obtaining average gray values of pixel points on the two actual edge lines, and carrying out negative correlation mapping and normalization on the absolute value of the difference value of the average gray values to obtain a gray value difference characteristic value;
and multiplying the distance distribution characteristic values and the gray value difference characteristic values of the two actual edge lines, wherein the product is used as the matching degree of the two actual edge lines.
Further, the obtaining of the optimized high threshold includes:
the calculation formula of the optimized high threshold value is as follows:
in the method, in the process of the invention,
expressed as end points
Is used for the optimization of the high threshold value of (c),
indicated as a preset high threshold value,
represented as actual edge lines
Is used for the total number of pixels of a display device,
represented as actual edge lines
Upper first
The maximum gray value difference of the individual pixels,
represented as actual edge lines
Upper first
The maximum gray value difference of the individual pixels,
represented as actual edge lines
Upper first
Each pixel point and end point
Is used for the distance of euclidean distance,
represented as a normalization function.
Further, the acquiring of the actual edge point and the new actual edge line includes:
screening preselected actual edge points adjacent to the endpoints in the unconnected area according to the optimized high threshold of the endpoints, and selecting another endpoint in the unconnected area for screening when the preselected actual edge points cannot be screened according to the optimized high threshold;
when only one preselected actual edge point exists, the preselected actual edge point is used as an actual edge point and is connected with the end point, and a new actual edge line is obtained;
when the preselected actual edge points are two or more, a first slope of each endpoint and adjacent pixel points on the actual edge line is obtained, second slopes of all the preselected actual edge points and the endpoint are calculated, and the corresponding preselected actual edge point with the smallest difference between the first slope and the second slope is used as an actual edge point to be connected with the endpoint, so that a new actual edge line is obtained;
the actual edge point is the end point of the new actual edge line.
Further, the screening the oil stain area according to the gray value change rule in the closed area includes:
obtaining a gray value entropy value of a pixel point in the closed region, carrying out normalization processing on the gray value entropy value to obtain a gray rule characteristic value, and marking the closed region as a defect region when the gray rule characteristic value is larger than a preset gray threshold value; and when the gray rule characteristic value is smaller than or equal to a preset gray value threshold value, marking the closed area as an oil pollution area.
Further, the step of screening out the abnormal point on the initial edge line through the abnormal point probability to obtain an actual edge line includes:
obtaining the abnormal point probability of each pixel point on the initial edge line, and screening out the corresponding pixel point as an abnormal point when the abnormal point probability is larger than a preset abnormal threshold value; and when the probability of the abnormal point is smaller than or equal to a preset abnormal threshold value, the corresponding pixel point is used as a point on the actual edge line, and the actual edge line is obtained.
Further, the step of screening out the actual edge line group formed by two similar actual edge lines according to the matching degree includes:
obtaining the matching degree of any two actual edge lines, and taking the corresponding two actual edge lines as a matching group when the matching degree is larger than a preset matching threshold value; when the matching degree is smaller than or equal to a preset matching threshold value, the corresponding two actual edge lines are continuously matched with other actual edge lines;
if one of the actual edge lines only has one matching group with the other of the actual edge lines, the matching group corresponding to the two actual edge lines is used as an actual edge line group; and if one actual edge line and two or more actual edge lines form two or more matching groups, taking the matching group with the largest matching degree corresponding to the matching group as an actual edge line group corresponding to the two actual edge lines.
The invention has the following beneficial effects:
according to the method, the characteristic analysis of the gray level characteristic and the spreading characteristic of the greasy dirt edge is carried out, the local edge is analyzed by the double-threshold selection part during the detection of the canny operator, the high threshold is optimized, and the local self-adaptive selection of the high threshold is realized. Firstly, the selection of an initial edge line is completed through a preset high threshold value, abnormal points on the initial edge line are screened out through the gray change difference and the relative distance between pixel points in the initial edge line, the abnormal points are generated because the selection of the preset high threshold value is too low, the obtained actual edge line is more real and accurate through screening out the abnormal points, and the subsequent matching result of the actual edge line is more accurate. Further, the matched actual edge line groups are obtained according to the position distribution and gray value difference of the actual edge lines, and the actual edge line groups are connected, and at the moment, the situation that edge points existing between the matched actual edge lines are not detected is caused by too high selection of a preset high threshold value. And analyzing gradient changes of adjacent pixel points on the actual edge line to be connected, adjusting a preset high threshold value to obtain an optimized high threshold value, wherein the optimized high threshold value is the optimal high threshold value corresponding to the end point of the actual edge line to be connected, and obtaining a new actual edge line by analyzing the pixel points by the optimized high threshold value to finish connection to obtain an actual edge line to be detected, wherein the actual edge line to be detected is an accurate edge line after connection. And (3) carrying out iterative matching connection to finally obtain a closed region and finish edge detection, wherein the problem of local optimal solution caused by single adjustment of the threshold value is avoided because the high threshold value is required to be updated and optimized in a self-adaptive mode when the actual edge point is searched in each iterative process. And obtaining an oil stain area according to the change rule of the gray value in the obtained closed area containing the continuous edge, so that the oil stain area is extracted more accurately and completely, the defect detection is more accurate, and the quality of the tab mold is improved.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a method for detecting a defect of oil stain on a surface of a tab mold according to an embodiment of the present invention.
Detailed Description
In order to further describe the technical means and effects adopted by the invention to achieve the preset aim, the following is a detailed description of specific implementation, structure, characteristics and effects of the defect detection method for the oil stain on the surface of the tab mold according to the invention with reference to the accompanying drawings and the preferred embodiment. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The following specifically describes a specific scheme of the defect detection method for the oil stain on the surface of the tab mold provided by the invention with reference to the accompanying drawings.
Referring to fig. 1, a flowchart of a method for detecting a defect of oil stain on a surface of a tab mold according to an embodiment of the invention is shown, and the method includes the steps of:
s1: obtaining a gray level image of the surface of the tab mold, and carrying out canny algorithm edge detection on the gray level image of the surface of the tab mold through a preset high threshold value to obtain an initial edge line; the abnormal point probability of each pixel point is obtained through the gray value change difference and the relative distance between each pixel point and other pixel points on the initial edge line, and the abnormal points on the initial edge line are screened out through the abnormal point probability, so that the actual edge line is obtained.
According to the invention, the greasy dirt area is extracted to finish preliminary defect detection mainly by carrying out canny operator edge detection on the surface of the tab die, and for the tab die, the greasy dirt defect on the smooth surface belongs to the surface defect which can be easily removed, and some scratches, bulges or holes and other defects are main defects which need further maintenance, so that during defect detection, the greasy dirt area can interfere with the identification of the main defects, and general greasy dirt is lower in contrast than background images, and is not easy to separate the greasy dirt from the background. Therefore, the gray change characteristics and the spreading characteristics of the oil stain area are analyzed, and the detection is performed by adopting an improved canny operator method, so that when the detection of the surface of the lug mold is performed, the gray image of the surface of the lug mold is required to be acquired firstly.
In the embodiment of the invention, an industrial camera is used, a fixed light source is adopted to collect an image of the surface of the lug mold, the obtained image of the surface of the lug mold is an RGB image, and a weighted graying method is adopted to carry out graying treatment on the image of the surface of the lug mold to obtain a gray image of the surface of the lug mold, wherein the weighted graying method is a technical means well known to a person skilled in the art, and the graying treatment also comprises an average value method, a maximum value method and the like, and is not limited herein.
And after the gray level image of the surface of the tab mold is obtained, carrying out canny algorithm edge detection on the gray level image of the surface of the tab mold through a preset high threshold value to obtain an initial edge line. According to experience, when the oil stain is polluted by contacting the surface of the mould, the oil stain can spread around the contact point, and the gray value change of the surface of the oil stain area can be different due to different depths and spreading trends of the oil stain, so that the gradient of the oil stain edge pixel points at different positions is also different, and when a preset high threshold value is used for canny operator detection, the spreading change characteristic of the oil stain is not considered, and the obtained edge pixel points are inaccurate. In the embodiment of the invention, the preset high threshold is set to be 50 according to experience, and a specific numerical value implementation can be adjusted according to specific situations.
Because the preset high threshold is set manually, abnormal edge points which are detected by mistake may exist on the initial edge lines, and part of real edge points are not detected among the initial edge lines, so that an area surrounded by the edge contour is not a closed area, namely defect information is lost, and a defect identification result is incomplete. Therefore, in order to obtain the real closed area, the initial edge line is firstly analyzed, abnormal points are screened out according to the gradient characteristics of the real edge points and the false edge points, all the real edge lines are obtained, and the real edge lines are conveniently connected to obtain the complete closed area. Therefore, the abnormal point probability of each pixel point is obtained through the gray value variation difference and the relative distance between each pixel point and other pixel points on the initial edge line, the abnormal point probability can reflect the difference between each pixel point and other pixel points on gradient information, and the abnormal point on the initial edge line is screened out through the abnormal point probability, so that the actual edge line is obtained, and the method specifically comprises the following steps:
the abnormal point is caused by the fact that the selection of a preset high threshold value is too low during the detection of the canny operator, so that the gray value of the abnormal point is obviously changed, but the difference between the gray value of the abnormal point and the pixel points on other normal edge lines is large, and the probability that each pixel point is the abnormal point can be judged through gradient information, namely the gray value change degree.
The method for acquiring the gradient information of the pixel point, namely the gray value change degree comprises the following steps: and acquiring the gray value of a neighborhood pixel in a preset neighborhood range of each pixel on the initial edge line, and taking the maximum difference value between the gray value of each pixel and the gray value of the neighborhood pixel as the maximum gray value difference, namely the gradient information of each pixel. In the embodiment of the invention, the preset neighborhood range is an eight-neighborhood range around the pixel point, and it is to be noted that in the prior art, the gradient value is calculated by combining gray value changes in multiple directions.
Taking one pixel point on the initial edge line as a reference pixel point, taking the reference pixel point as an example, and solving the probability that each pixel point is an abnormal point. And calculating the maximum gray value difference between the reference pixel point and other pixel points on an initial edge line corresponding to the reference pixel point, and taking the absolute value of the difference value of the maximum gray value difference as the gradient difference. And calculating and normalizing the absolute value of the gray value difference between the reference pixel point and other pixel points to obtain the reference gray value difference.
Meanwhile, the influence is different in consideration of the fact that the relative positions of other pixel points and the reference pixel are different. Therefore, euclidean distance is used as weight, the Euclidean distance between the reference pixel point and other pixel points is mapped in a negative correlation mode and normalized, adjustment weight is obtained, and when the other pixel points are closer to the reference pixel point, the difference influence is larger. It should be noted that the euclidean distance is a technical means well known to those skilled in the art, and will not be described herein.
And multiplying the reference gray value difference by the gradient difference to obtain a gray value change difference, obtaining the gray value change difference after weighting the reference pixel point and each other pixel point by adjusting the weight, and obtaining the average value of all the weighted gray value change differences to obtain the probability of the abnormal point. The abnormal point probability reflects the similarity degree of the reference pixel point and other pixel points in gray level change and relative positions, and when the reference pixel point is dissimilar from the other pixel points, namely the difference is larger, the abnormal point probability is larger. In the embodiment of the invention, for the accuracy of subsequent calculation, the outlier probability formula is:
in the method, in the process of the invention,
represented as reference pixel points
Is used for the probability of an outlier,
represented as reference pixel points 
The total number of pixels corresponding to the initial edge line,
represented as reference pixel points
Is used for the gray-scale value of (c),
represented as reference pixel points
Corresponding to the first edge line
The gray values of the other pixel points,
represented as reference pixel points
And the corresponding initial edge line
Euclidean distances between other pixels,
represented as reference pixel points
The maximum gray value difference of (a), i.e., gradient information;
represented as reference pixel points
Corresponding to the first edge line
Maximum gray value difference of other pixels, namely gradient information;
it should be noted that, normalization is a technical means well known to those skilled in the art, and the normalization function may be selected by linear normalization or standard normalization, and the specific normalization method is not limited herein.
Analyzing the variation difference of the integral gray value of the reference pixel point and other pixel points by adopting an average value method,
expressed as the absolute value of the difference in gray values,
the method is represented by carrying out normalization processing on the absolute value of the gray value difference to obtain a reference gray value difference, when the reference gray value difference is larger, the reference pixel point and other pixel points are more likely not to be on the same edge, so that the probability that the reference pixel point is an abnormal point is larger, 
The gradient difference between the reference pixel point and other pixel points is expressed, and when the gradient difference is larger, the degree of change of the reference pixel point and other pixel points is not consistent, so that the probability that the reference pixel point is an abnormal point is larger. The reference gray value difference and the gradient difference are in positive correlation with the probability of the abnormal point.
Adding Euclidean distance as adjusting weight, considering influence of relative position of other pixels on reference pixel point, and influence degree of other pixels close to the reference pixel point is larger, so that
The adjustment weight obtained by performing negative correlation mapping and normalization on the Euclidean distance is represented as the closer the distance is, the larger the influence of other pixel points is, so the larger the weight is given. At the moment, after the position relation is considered, the average value of all gray level change differences is calculated, and the obtained abnormal point probability can represent the gray level value change difference between the reference pixel point and other pixel points.
When the probability of the abnormal point of the reference pixel point is larger, the larger the gray value change difference between the reference pixel point and other pixel points is, and further the larger the probability that the reference pixel point does not accord with the edge characteristic of the corresponding initial edge line and is the false detection edge point is. Therefore, the probability of abnormal point of each pixel point on the initial edge line is obtained, and in the embodiment of the invention, the preset abnormal threshold value is set to be 0.8. When the probability of the abnormal point is larger than a preset abnormal threshold, the corresponding pixel point is larger in difference with other edge pixel points, and the corresponding pixel point is taken as the abnormal point to be screened out; when the probability of the abnormal point is smaller than or equal to a preset abnormal threshold value, the corresponding pixel point is smaller in difference with other edge pixel points, and the corresponding pixel point is taken as a point on an actual edge line.
After the abnormal point is screened out, at least one actual edge line is obtained. Because, when the continuity of the initial edge line is not affected after the outlier is screened out, the obtained actual edge line is one, and when the continuity of the initial edge line is affected by the screening out of the outlier, the initial edge line is divided into a plurality of actual edge lines, and at this time, the actual edge line analysis needs to be connected subsequently.
Thus, the preliminary screening of the edge line is completed, and abnormal points generated due to the fact that the preset high threshold value is too low are screened out, so that the more accurate actual edge line is obtained.
S2: and obtaining matching degree through gray value difference and position distribution between the actual edge lines, and screening out an actual edge line group formed by two similar actual edge lines according to the matching degree.
After the actual edge line is obtained, because the edge lines detected by the canny operator are segmented edge lines, the situation that the actual edge is not detected due to the fact that a preset high threshold value is too high exists between the segmented edge lines, and in order to accurately extract the oil stain area, an accurate closed area needs to be obtained. Therefore, it is further necessary to connect all edge lines to obtain a closed area.
When the edge line connection is carried out, whether two edges are corresponding to one edge or not needs to be judged, after screening according to S1, a plurality of actual edge lines can be obtained, gray value differences and position distribution among the actual edge lines are analyzed, and matching degree is obtained. The matching degree can represent the similarity degree of two actual edge lines, and the more similar actual edge lines are, the more likely to correspond to the same edge line, and the higher the probability that a real edge point exists between the two similar actual edge lines. Therefore, the method for obtaining the specific matching degree comprises the following steps:
and obtaining average gray values of pixel points on the two actual edge lines, carrying out negative correlation mapping and normalization on the absolute value of the difference value of the average gray values, and obtaining the gray value difference characteristic value. The gray value difference characteristic value reflects the gray similarity of the actual edge lines, and the larger the gray value difference characteristic value is, the larger the matching degree is.
And obtaining the end points of each actual edge line, carrying out negative correlation mapping and normalization processing on the minimum value of Euclidean distance between the end points of the two actual edge lines, and obtaining the distance distribution characteristic value. The distance distribution characteristic value reflects the position distribution characteristics of the two actual edge lines, and when the distance between the two actual edge lines is closer, the corresponding distance distribution characteristic value is larger, and the corresponding matching degree is also larger.
And multiplying the distance distribution characteristic value and the gray value difference characteristic value of the two actual edge lines, and taking the product as the matching degree of the two actual edge lines. The probability of the existence of the real edge point between the two real edge lines can be reflected through the matching degree, and the two real edge lines with larger matching degree correspond to the two real edge lines with the more possibility of the existence of the real edge point. In the embodiment of the invention, for the accuracy of subsequent calculation, a specific matching degree formula is as follows:
in the method, in the process of the invention,
represented as actual edge lines
And the actual edge line
The degree of matching between the two,
represented as actual edge lines
And the actual edge line
The euclidean distance between the two,
represented as actual edge lines
Is used for the color filter,
represented as actual edge lines
Is used for the color filter,
represented as a minimum value extraction function,
represented by a natural constant which is a function of the natural constant,
represented as a normalization function.
Combining gray value difference characteristic values and distance distribution through the form of productsThe characteristic value is analyzed and the characteristic value is obtained,
expressed as the minimum of the euclidean distance, i.e. the minimum distance between the two actual edge lines.
The method is expressed as that the Euclidean distance minimum value is subjected to negative correlation mapping and normalization processing to obtain a distance distribution characteristic value, and when the distance distribution characteristic value is larger, the smaller the minimum distance between two actual edge lines is, the greater the matching degree is. 
Expressed as the absolute value of the mean difference between two actual edge lines,
the method is characterized in that an exponential function based on a natural constant is adopted for carrying out negative correlation mapping and normalization, a gray value difference characteristic value is obtained, and when the gray value difference characteristic value is larger, the smaller the average gray value difference between two actual edge lines is, the larger the matching degree is. The gray value difference characteristic value and the distance distribution characteristic value are in positive correlation with the matching degree.
According to the matching degree, an actual edge group formed by two similar actual edge lines can be screened, and the specific grouping process is as follows: obtaining the matching degree of any two actual edge lines, when the matching degree is larger than a preset matching threshold value, indicating that undetected actual edge points exist between the two actual edge lines, and taking the corresponding two actual edge lines as a matching group; when the matching degree is smaller than or equal to a preset matching threshold, it is indicated that there may be no actual edge point between the two actual edge lines, and the corresponding two actual edge lines are continuously matched with other actual edge lines.
In the matching process, one matching group corresponds to two actual edge lines, and when one matching group exists between one actual edge line and the other actual edge line, the matching group corresponding to the two actual edge lines is taken as an actual edge line group. When there is one actual edge line and two or more actual edge lines to form two or more matching groups, the matching group with the largest corresponding matching degree in the matching groups is taken as the actual edge line group corresponding to the two actual edge lines. The two most similar actual edge lines are found, one actual edge line corresponding to at most one actual edge line group.
And finally, completing primary matching on all the actual edge lines to obtain an actual edge line group, and further connecting two actual edge lines in the matched actual edge line group.
S3: taking any actual edge line group as a detection group, and starting from an endpoint of one actual edge line in the detection group in an unconnected area between the actual edge lines in the detection group, adjusting a preset high threshold according to gradient changes of adjacent pixel points on the actual edge line to obtain an optimized high threshold of the endpoint; screening by optimizing a high threshold value to obtain an actual edge point adjacent to the end point in the unconnected area, and obtaining a new actual edge line; starting from the actual edge points of the new actual edge lines, continuously adjusting a preset high threshold value to obtain all the actual edge points, and completing connection to obtain the actual edge lines to be detected of the detection group; and obtaining the actual edge lines to be detected corresponding to each actual edge line group.
In the step S2, a matched actual edge line set may be obtained, in which an undetected actual edge point exists between two actual edge lines, which indicates that the preset high threshold is too large, and the preset high threshold needs to be reduced and optimized, and gradient information of the edge line is mainly analyzed at the actual edge line end point, and the local self-adaptive high threshold at the end point, that is, the optimized high threshold, is obtained by optimizing the preset high threshold through the gradient information of the actual edge line. The real edge points can be further screened out for connection by optimizing the high threshold. Therefore, an arbitrary actual edge line group is first used as a detection group, and the connection process is performed according to the detection group.
In the detection group, the area to be connected between the two actual edge lines is taken as an unconnected area, and in the embodiment of the invention, the specific unconnected area selection method comprises the following steps: and connecting two end points of the minimum Euclidean distance corresponding to the two actual edge line end points with the two minimum end points to obtain a preset edge line, taking the pixel point on the preset edge line as the center, obtaining a neighborhood pixel point of the 5*5 size of each pixel point, and taking the pixel points on all the preset edge line and the area formed by the corresponding neighborhood pixel points as the unconnected area. In the unconnected area, the acquisition of the actual edge points is performed.
Starting from an endpoint of an actual edge line in the detection group, namely starting from a minimum endpoint in the unconnected area, adjusting a preset high threshold according to gradient change of adjacent pixel points on the actual edge line, and obtaining an optimized high threshold of the endpoint, wherein a calculation formula of the optimized high threshold is as follows:
in the method, in the process of the invention,
expressed as end points
Is used for the optimization of the high threshold value of (c),
indicated as a preset high threshold value,
represented as actual edge lines
Is used for the total number of pixels of a display device,
represented as actual edge lines
Upper first
The maximum gray value difference of the individual pixels, 
Represented as actual edge lines
Upper first
The maximum gray value difference of the individual pixels,
represented as actual edge lines
Upper first
Each pixel point and end point
Is used for the distance of euclidean distance,
represented as a normalization function.
The preset high threshold is lowered by the mean value of the gradient change,
represented as actual edge lines
Except for the end points
Is used for the total number of pixels of a display device,
the absolute value of the difference value expressed as the maximum gray value difference between two adjacent pixel points is recorded as the gradient change difference, and if the gradient change difference between the pixel points is larger, the degree of the preset high threshold value to be adjusted is larger, and the finally obtained optimized high threshold value is smaller.
Denoted as the first
Each pixel point and endPoint(s)
Performing negative correlation mapping and normalization processing on Euclidean distance of (2) to obtain position weights, adjusting the influence of gradient variation differences at different positions through the position weights, and when the distance is equal to the end point
The closer the gradient change difference effect of adjacent pixel points is, the larger.
Wherein the method comprises the steps of
Represented as the mean of the gradient change differences adjusted according to the position weights, noted as the overall gradient change difference, which may be reflected at the endpoints
Where the actual edge line
The relative gradient change degree is obtained at the end point
An optimized high threshold. The higher the overall gradient change difference, the lower the optimized high threshold.
In the unconnected area, the pixel points can be screened at the end points by optimizing a high threshold value, so as to obtain actual edge points adjacent to the end points in the unconnected area, and a new actual edge line is obtained, wherein two end points which are the minimum end points corresponding to the two actual edge lines respectively exist in the unconnected area, and the specific acquisition process comprises the following steps:
and screening the pixel points adjacent to the end points in the unconnected area according to the optimized high threshold value at the end points, namely the optimized high threshold value at the minimum end point, taking the pixel points higher than the optimized high threshold value as preselected actual edge points, and selecting to start screening from the other end point in the unconnected area, namely the other minimum end point if the preselected actual edge points cannot be screened according to the optimized high threshold value. It should be noted that, because the possibility of the edge points existing between the actual edge lines is already determined in the matching process of the actual edge lines, the possibility that both end points cannot be screened is extremely low, and if both end points cannot be screened, the two actual edge lines are considered to be connected as one edge line.
When only one preselected actual edge point exists, the preselected actual edge point is the actual edge point connected with the end point, the preselected actual edge point is used as the actual edge point and is connected with the end point, and a new actual edge line is obtained.
When more than two preselected actual edge points exist, the fact that a plurality of pixel points conforming to the actual edge points exist is indicated, the actual edge points which best conform to the actual edge line are required to be further selected, the slopes of the adjacent pixel points on the endpoint and the actual edge line are calculated to be used as a first slope, and the slopes of all the preselected actual edge points and the endpoint are calculated to be used as a second slope. The slope may reflect the trend of the local edge, and when the trend of the local edge is similar, the corresponding preselected actual edge point is the edge point in the real edge.
Calculating the absolute value of the difference between the first slope and each second slope as a slope difference, wherein the slope difference reflects the difference degree of the edge change trend, taking a preselected actual edge point corresponding to the minimum slope difference as an actual edge point, enabling the actual edge point to be more in line with the change trend of the actual edge at the moment, and connecting the actual edge point with an endpoint to obtain a new actual edge line.
The obtained actual edge point is the end point of the new actual edge line, further, the end point of the new actual edge line can be continuously analyzed, namely, the actual edge point is continuously changed by a new gradient to obtain a new optimized high threshold value, the new actual edge point is continuously obtained, and finally, the connection of the two actual edge lines is completed. And taking the connected edge lines as actual edge lines to be detected of the detection groups, and obtaining corresponding actual edge lines to be detected for each actual edge line group based on the connection process of the detection groups.
So far, one-time connection of each actual edge group is completed, and the connected actual edge line to be detected is obtained.
S4: iteratively matching and connecting all the actual edge lines to be detected until a closed area is obtained; and screening out the oil-polluted area according to the gray value change rule in the closed area.
And (3) based on the matching process in the step (S2), matching all the actual edge lines to be detected again to obtain a new actual edge group, continuing to connect through the connecting process in the step (S3), and continuously iterating to perform matching connection until a complete closed area is obtained. It should be noted that, in the process of matching again, the actual edge lines to be matched also include the remaining unconnected actual edge lines, so as to ensure that the edge lines similar to the matching to the greatest extent obtain more complete closed areas. And when new actual edge points are continuously and iteratively found, the high threshold value is continuously and adaptively updated and optimized, so that the problem of local optimal solution caused by single adjustment of the threshold value is avoided.
In the gray scale image of the mold surface, a plurality of closed areas may be obtained, which contain not only greasy dirt areas but also defective areas, thus further distinguishing the closed areas. According to the oil pollution area is an area formed by spreading oil drops on the surface, the surface of the oil pollution area has certain regularity, and the surface of the defect area is often irregular, so that the oil pollution area can be screened out according to the gray level change rule of the surface of the closed area, and the method specifically comprises the following steps:
Obtaining a gray value entropy value of a pixel point in the closed area, carrying out normalization processing on the gray value entropy value to obtain a gray rule characteristic value, and reflecting the gray change rule degree through the gray rule characteristic value. When the gray rule characteristic value is larger than a preset gray threshold value, the gray value change in the closed area is disordered, and the probability that the closed area is an oil stain area is extremely low, so that the corresponding closed area is marked as a defect area.
When the gray rule characteristic value is smaller than or equal to a preset gray value threshold value, the gray value change of the pixel points in the closed area is regular, and the corresponding closed area can be marked as an oil pollution area. In the embodiment of the present invention, the preset gray value threshold is 0.5, and the specific numerical value implementation can be adjusted according to the specific implementation condition, and it should be noted that the normalization processing and the calculation of the entropy value are all technical means well known to those skilled in the art, and are not described herein.
And extracting the greasy dirt region in the gray level image on the surface of the tab so as to finish the detection of the greasy dirt defect.
In summary, the method includes the steps of firstly obtaining a gray level image of the surface of a tab mold, detecting by using a preset canny operator high threshold to obtain initial edge lines, screening abnormal points through gray level variation differences and relative positions of the initial edge lines to obtain actual edge lines, matching according to position distribution and gray level differences of the actual edge lines to obtain an actual edge line group, adjusting the preset high threshold according to gradient variation on the actual edge lines in an unconnected area of the actual edge line group to obtain an optimized high threshold at an end point, screening out actual edge points to be connected with the end point according to the optimized high threshold to obtain new actual edge lines, continuously adjusting the preset high threshold through the new actual edge lines to obtain all actual edge points to be detected to finish connection, obtaining the actual edge lines to be detected of each actual edge group, iteratively matching all the actual edge lines to be detected until a complete closed area is obtained, and screening out an oil stain area according to a change rule of gray level values in the closed area. According to the invention, through image processing, the high threshold value is subjected to local self-adaptive optimization through gradient information and spatial information, so that a more complete and accurate oil stain region is obtained.
It should be noted that: the sequence of the embodiments of the present invention is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments.
Claims (8)
1. The defect detection method for the oil stain on the surface of the lug mold is characterized by comprising the following steps of:
obtaining a gray level image of the surface of a tab mold, and carrying out canny algorithm edge detection on the gray level image of the surface of the tab mold through a preset high threshold value to obtain an initial edge line; obtaining abnormal point probability of each pixel point according to gray value variation difference and relative distance between each pixel point and other pixel points on the initial edge line, and screening abnormal points on the initial edge line according to the abnormal point probability to obtain an actual edge line;
obtaining matching degree through gray value difference and position distribution between the actual edge lines, and screening out an actual edge line group formed by two similar actual edge lines according to the matching degree;
Taking any actual edge line group as a detection group, and in an unconnected area between the actual edge lines in the detection group, starting from an endpoint of one actual edge line in the detection group, adjusting a preset high threshold according to gradient change of adjacent pixel points on the actual edge line to obtain an optimized high threshold of the endpoint; screening through the optimized high threshold value to obtain an actual edge point adjacent to the end point in the unconnected area, and obtaining a new actual edge line; starting from the actual edge points of the new actual edge lines, continuously adjusting a preset high threshold value to obtain all the actual edge points, and completing connection to obtain the actual edge lines to be detected of a detection group; obtaining the actual edge lines to be detected corresponding to each actual edge line group;
iteratively matching and connecting all the actual edge lines to be detected until a closed area is obtained; and screening out the oil-polluted area according to the gray value change rule in the closed area.
2. The method for detecting the defect of the oil stain on the surface of the tab mold according to claim 1, wherein the obtaining of the abnormal point probability of each pixel point comprises the following steps:
Obtaining the maximum gray value difference between each pixel point on the initial edge line and the neighborhood pixel point in a preset neighborhood range;
taking one pixel point on the initial edge line as a reference pixel point, calculating and normalizing gray value differences between the reference pixel point and other pixel points on the initial edge line to obtain reference gray value differences;
taking the difference of the maximum gray value difference between the reference pixel point and other pixel points on the initial edge line as a gradient difference; multiplying the reference gray value difference by the gradient difference, and taking the product as a gray value change difference;
performing negative correlation mapping and normalization on Euclidean distances between the reference pixel point and other pixel points on the initial edge line to obtain an adjustment weight; and obtaining the gray value variation difference between the reference pixel point and all other pixel points, adjusting the gray value variation difference through the adjusting weight, and taking the average value of all the weighted gray value variation differences as the probability of the abnormal point.
3. The method for detecting the defect of the oil stain on the surface of the tab mold according to claim 1, wherein the obtaining of the matching degree comprises the following steps:
Acquiring the end points of each actual edge line, performing negative correlation mapping and normalization processing on the minimum value of Euclidean distance between the two actual edge line end points, and acquiring a distance distribution characteristic value;
obtaining average gray values of pixel points on the two actual edge lines, and carrying out negative correlation mapping and normalization on the absolute value of the difference value of the average gray values to obtain a gray value difference characteristic value;
and multiplying the distance distribution characteristic values and the gray value difference characteristic values of the two actual edge lines, wherein the product is used as the matching degree of the two actual edge lines.
4. The method for detecting the defect of the oil stain on the surface of the tab mold according to claim 2, wherein the obtaining of the optimized high threshold value comprises:
the calculation formula of the optimized high threshold value is as follows:
in the method, in the process of the invention,
expressed as endpoint->
Is optimized high threshold,/-, is preferred>
Expressed as preset high threshold,/->
Represented as the actual edge line +.>
Total number of pixels, < >>
Represented as the actual edge line +.>
Go up to->
Maximum gray value difference of individual pixels,/->
Represented as the actual edge line +.>
Go up to->
Maximum gray value difference of individual pixels,/->
Represented as the actual edge line +.>
Go up to->
Pixels and end points- >
Euclidean distance of>
Represented as a normalization function.
5. The method for detecting the defect of the oil stain on the surface of the tab mold according to claim 1, wherein the obtaining of the actual edge point and the new actual edge line comprises the following steps:
screening preselected actual edge points adjacent to the endpoints in the unconnected area according to the optimized high threshold of the endpoints, and selecting another endpoint in the unconnected area for screening when the preselected actual edge points cannot be screened according to the optimized high threshold;
when only one preselected actual edge point exists, the preselected actual edge point is used as an actual edge point and is connected with the end point, and a new actual edge line is obtained;
when the preselected actual edge points are two or more, a first slope of each endpoint and adjacent pixel points on the actual edge line is obtained, second slopes of all the preselected actual edge points and the endpoint are calculated, and the corresponding preselected actual edge point with the smallest difference between the first slope and the second slope is used as an actual edge point to be connected with the endpoint, so that a new actual edge line is obtained;
the actual edge point is the end point of the new actual edge line.
6. The method for detecting the oil stain defect on the surface of the tab mold according to claim 1, wherein the step of screening the oil stain region according to the gray value change rule in the closed region comprises the following steps:
obtaining a gray value entropy value of a pixel point in the closed region, carrying out normalization processing on the gray value entropy value to obtain a gray rule characteristic value, and marking the closed region as a defect region when the gray rule characteristic value is larger than a preset gray threshold value; and when the gray rule characteristic value is smaller than or equal to a preset gray value threshold value, marking the closed area as an oil pollution area.
7. The method for detecting the defect of the oil stain on the surface of the tab mold according to claim 1, wherein the step of screening out the abnormal point on the initial edge line by the abnormal point probability to obtain an actual edge line comprises the following steps:
obtaining the abnormal point probability of each pixel point on the initial edge line, and screening out the corresponding pixel point as an abnormal point when the abnormal point probability is larger than a preset abnormal threshold value; and when the probability of the abnormal point is smaller than or equal to a preset abnormal threshold value, the corresponding pixel point is used as a point on the actual edge line, and the actual edge line is obtained.
8. The method for detecting the oil stain defect on the surface of the tab mold according to claim 1, wherein the step of screening out the actual edge line group consisting of two similar actual edge lines according to the matching degree comprises the steps of:
obtaining the matching degree of any two actual edge lines, and taking the corresponding two actual edge lines as a matching group when the matching degree is larger than a preset matching threshold value; when the matching degree is smaller than or equal to a preset matching threshold value, the corresponding two actual edge lines are continuously matched with other actual edge lines;
if one of the actual edge lines only has one matching group with the other of the actual edge lines, the matching group corresponding to the two actual edge lines is used as an actual edge line group; and if one actual edge line and two or more actual edge lines form two or more matching groups, taking the matching group with the largest matching degree corresponding to the matching group as an actual edge line group corresponding to the two actual edge lines.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310443245.0A CN116309537B (en) | 2023-04-24 | 2023-04-24 | Defect detection method for oil stain on surface of tab die |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310443245.0A CN116309537B (en) | 2023-04-24 | 2023-04-24 | Defect detection method for oil stain on surface of tab die |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116309537A true CN116309537A (en) | 2023-06-23 |
| CN116309537B CN116309537B (en) | 2023-08-01 |
Family
ID=86780019
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310443245.0A Active CN116309537B (en) | 2023-04-24 | 2023-04-24 | Defect detection method for oil stain on surface of tab die |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116309537B (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116612123A (en) * | 2023-07-21 | 2023-08-18 | 山东金胜粮油食品有限公司 | Visual detection method for peanut oil processing quality |
| CN116664584A (en) * | 2023-08-02 | 2023-08-29 | 东莞市旺佳五金制品有限公司 | Intelligent feedback regulating system for production of thin-wall zinc alloy die casting die |
| CN116703892A (en) * | 2023-08-01 | 2023-09-05 | 东莞市京品精密模具有限公司 | Image data-based lithium battery cutter abrasion evaluation and early warning method |
| CN116758071A (en) * | 2023-08-17 | 2023-09-15 | 青岛冠宝林活性炭有限公司 | Intelligent detection method for carbon electrode dirt under visual assistance |
| CN116823814A (en) * | 2023-08-28 | 2023-09-29 | 山东巨鑫伟业电气设备有限公司 | Distribution equipment welding production defect detection system |
| CN117011297A (en) * | 2023-10-07 | 2023-11-07 | 惠州市凯默金属制品有限公司 | Aluminum alloy automobile accessory die defect detection method based on image processing |
| CN117291985A (en) * | 2023-11-24 | 2023-12-26 | 青岛宜霖赛瑞智能科技有限公司 | Image positioning method for part punching |
| CN117455870A (en) * | 2023-10-30 | 2024-01-26 | 太康精密（中山）有限公司 | Connecting wire and connector quality visual detection method |
| CN117723548A (en) * | 2023-12-14 | 2024-03-19 | 东莞市毅廷音响科技有限公司 | Automobile horn production quality detection method |
| CN118014998A (en) * | 2024-04-09 | 2024-05-10 | 深圳市富安娜艺术家居有限公司 | Bed sheet cloth detection method and system based on machine vision |
| CN118014998B (en) * | 2024-04-09 | 2024-06-11 | 深圳市富安娜艺术家居有限公司 | Bed sheet cloth detection method and system based on machine vision |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2009113231A1 (en) * | 2008-03-14 | 2009-09-17 | 株式会社ソニー・コンピュータエンタテインメント | Image processing device and image processing method |
| CN106204453A (en) * | 2016-07-14 | 2016-12-07 | 京东方科技集团股份有限公司 | The interpolation method of a kind of image and device |
| CN115272341A (en) * | 2022-09-29 | 2022-11-01 | 华联机械集团有限公司 | Packaging machine defect product detection method based on machine vision |
| CN115272305A (en) * | 2022-09-26 | 2022-11-01 | 南通润一轩纺织品有限公司 | Button hole defect detection method |
-
2023
- 2023-04-24 CN CN202310443245.0A patent/CN116309537B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2009113231A1 (en) * | 2008-03-14 | 2009-09-17 | 株式会社ソニー・コンピュータエンタテインメント | Image processing device and image processing method |
| CN106204453A (en) * | 2016-07-14 | 2016-12-07 | 京东方科技集团股份有限公司 | The interpolation method of a kind of image and device |
| CN115272305A (en) * | 2022-09-26 | 2022-11-01 | 南通润一轩纺织品有限公司 | Button hole defect detection method |
| CN115272341A (en) * | 2022-09-29 | 2022-11-01 | 华联机械集团有限公司 | Packaging machine defect product detection method based on machine vision |
Non-Patent Citations (1)
| Title |
|---|
| 刘志赢 等: "基于侧抑制网络的改进Canny边缘检测算法", 舰船电子工程, vol. 40, no. 11, pages 48 - 52 * |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116612123A (en) * | 2023-07-21 | 2023-08-18 | 山东金胜粮油食品有限公司 | Visual detection method for peanut oil processing quality |
| CN116612123B (en) * | 2023-07-21 | 2023-10-13 | 山东金胜粮油食品有限公司 | Visual detection method for peanut oil processing quality |
| CN116703892B (en) * | 2023-08-01 | 2023-11-14 | 东莞市京品精密模具有限公司 | Image data-based lithium battery cutter abrasion evaluation and early warning method |
| CN116703892A (en) * | 2023-08-01 | 2023-09-05 | 东莞市京品精密模具有限公司 | Image data-based lithium battery cutter abrasion evaluation and early warning method |
| CN116664584A (en) * | 2023-08-02 | 2023-08-29 | 东莞市旺佳五金制品有限公司 | Intelligent feedback regulating system for production of thin-wall zinc alloy die casting die |
| CN116664584B (en) * | 2023-08-02 | 2023-11-28 | 东莞市旺佳五金制品有限公司 | Intelligent feedback regulating system for production of thin-wall zinc alloy die casting die |
| CN116758071A (en) * | 2023-08-17 | 2023-09-15 | 青岛冠宝林活性炭有限公司 | Intelligent detection method for carbon electrode dirt under visual assistance |
| CN116758071B (en) * | 2023-08-17 | 2023-11-03 | 青岛冠宝林活性炭有限公司 | Intelligent detection method for carbon electrode dirt under visual assistance |
| CN116823814A (en) * | 2023-08-28 | 2023-09-29 | 山东巨鑫伟业电气设备有限公司 | Distribution equipment welding production defect detection system |
| CN116823814B (en) * | 2023-08-28 | 2023-11-24 | 山东巨鑫伟业电气设备有限公司 | Distribution equipment welding production defect detection system |
| CN117011297A (en) * | 2023-10-07 | 2023-11-07 | 惠州市凯默金属制品有限公司 | Aluminum alloy automobile accessory die defect detection method based on image processing |
| CN117011297B (en) * | 2023-10-07 | 2024-02-02 | 惠州市凯默金属制品有限公司 | Aluminum alloy automobile accessory die defect detection method based on image processing |
| CN117455870A (en) * | 2023-10-30 | 2024-01-26 | 太康精密（中山）有限公司 | Connecting wire and connector quality visual detection method |
| CN117455870B (en) * | 2023-10-30 | 2024-04-16 | 太康精密（中山）有限公司 | Connecting wire and connector quality visual detection method |
| CN117291985A (en) * | 2023-11-24 | 2023-12-26 | 青岛宜霖赛瑞智能科技有限公司 | Image positioning method for part punching |
| CN117291985B (en) * | 2023-11-24 | 2024-03-15 | 青岛宜霖赛瑞智能科技有限公司 | Image positioning method for part punching |
| CN117723548A (en) * | 2023-12-14 | 2024-03-19 | 东莞市毅廷音响科技有限公司 | Automobile horn production quality detection method |
| CN118014998A (en) * | 2024-04-09 | 2024-05-10 | 深圳市富安娜艺术家居有限公司 | Bed sheet cloth detection method and system based on machine vision |
| CN118014998B (en) * | 2024-04-09 | 2024-06-11 | 深圳市富安娜艺术家居有限公司 | Bed sheet cloth detection method and system based on machine vision |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116309537B (en) | 2023-08-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116309537B (en) | Defect detection method for oil stain on surface of tab die | |
| CN113689428B (en) | Mechanical part stress corrosion detection method and system based on image processing | |
| CN115829883B (en) | Surface image denoising method for special-shaped metal structural member | |
| CN115619793B (en) | Power adapter appearance quality detection method based on computer vision | |
| CN108711158B (en) | Pointer instrument image identification method based on contour fitting and radial segmentation | |
| CN115457041A (en) | Road quality identification and detection method | |
| CN115345885A (en) | Method for detecting appearance quality of metal fitness equipment | |
| CN115147409B (en) | Mobile phone shell production quality detection method based on machine vision | |
| CN115294113A (en) | Wood veneer quality detection method | |
| CN116092013B (en) | Dangerous road condition identification method for intelligent monitoring | |
| CN114782329A (en) | Bearing defect damage degree evaluation method and system based on image processing | |
| CN116137036B (en) | Gene detection data intelligent processing system based on machine learning | |
| CN115082451B (en) | Stainless steel soup ladle defect detection method based on image processing | |
| CN116883408B (en) | Integrating instrument shell defect detection method based on artificial intelligence | |
| CN116309599B (en) | Water quality visual monitoring method based on sewage pretreatment | |
| CN116402810B (en) | Image processing-based lubricating oil anti-abrasive particle quality detection method | |
| CN115082477B (en) | Semiconductor wafer processing quality detection method based on light reflection removing effect | |
| CN116703909B (en) | Intelligent detection method for production quality of power adapter | |
| CN116051543B (en) | Defect identification method for peeling steel | |
| CN114998311A (en) | Part precision detection method based on homomorphic filtering | |
| CN116740053B (en) | Management system of intelligent forging processing production line | |
| CN117078688B (en) | Surface defect identification method for strong-magnetic neodymium-iron-boron magnet | |
| CN116612306A (en) | Computer vision-based intelligent flange plate alignment method and system | |
| CN116432052A (en) | Quality detection method for clamp for new energy automobile die | |
| CN114742849B (en) | Leveling instrument distance measuring method based on image enhancement |
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 |