CN117218114B - Mesh fabric defect rapid detection method based on image data processing - Google Patents
Mesh fabric defect rapid detection method based on image data processing Download PDFInfo
- Publication number
- CN117218114B CN117218114B CN202311462529.0A CN202311462529A CN117218114B CN 117218114 B CN117218114 B CN 117218114B CN 202311462529 A CN202311462529 A CN 202311462529A CN 117218114 B CN117218114 B CN 117218114B
- Authority
- CN
- China
- Prior art keywords
- defect
- mesh fabric
- distortion
- detection
- mesh
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 230000007547 defect Effects 0.000 title claims abstract description 174
- 239000004744 fabric Substances 0.000 title claims abstract description 129
- 238000001514 detection method Methods 0.000 title claims abstract description 87
- 238000012545 processing Methods 0.000 title claims abstract description 26
- 230000002950 deficient Effects 0.000 claims abstract description 32
- 238000005516 engineering process Methods 0.000 claims abstract description 31
- 238000000034 method Methods 0.000 claims abstract description 24
- 238000012937 correction Methods 0.000 claims abstract description 15
- 238000004364 calculation method Methods 0.000 claims abstract description 14
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 11
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 238000004519 manufacturing process Methods 0.000 claims description 18
- 238000009940 knitting Methods 0.000 claims description 10
- 210000002268 wool Anatomy 0.000 claims description 10
- 239000004753 textile Substances 0.000 claims description 9
- 238000009987 spinning Methods 0.000 claims description 6
- 230000009466 transformation Effects 0.000 claims description 6
- 238000009499 grossing Methods 0.000 claims description 5
- 238000004458 analytical method Methods 0.000 claims description 4
- 238000012163 sequencing technique Methods 0.000 claims description 4
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 230000006870 function Effects 0.000 claims description 3
- 238000005065 mining Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000012549 training Methods 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 abstract description 7
- 239000012467 final product Substances 0.000 abstract description 5
- 238000003908 quality control method Methods 0.000 description 4
- 230000004075 alteration Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 238000011179 visual inspection Methods 0.000 description 2
- 238000005034 decoration Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Landscapes
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
- Treatment Of Fiber Materials (AREA)
Abstract
The invention discloses a mesh fabric defect rapid detection method based on image data processing, which relates to the technical field of mesh fabric detection, and the method adopts image acquisition equipment and convolutional neural network technology, so that the defect on the surface of the mesh fabric can be detected efficiently and rapidly without consuming a large amount of manpower and time; the quality and the analyzability of the image are effectively improved through preprocessing sliding and geometric correction multiple technologies; the image frame is uniformly divided into a plurality of netlike detection areas, so that each area can be used for detecting and evaluating defects independently, and the parallelism and efficiency of detection are improved; through deep calculation and learning, comprehensively considering evaluation indexes of a plurality of factors of the defect area, and finally obtaining a distortion density indexThe method comprises the steps of carrying out a first treatment on the surface of the In a word, compared with the prior art, the method further reduces labor cost and defective rate, and further improves the quality of the net fabric by accurately detecting and evaluating defects, so that the final product meets the quality requirement of high standards.
Description
Technical Field
The invention relates to the technical field of mesh fabric detection, in particular to a mesh fabric defect rapid detection method based on image data processing.
Background
In the field of mesh fabrics based on image data processing, the production and quality control of mesh fabrics requires efficient and accurate methods to detect and evaluate defects thereof, including problems of holes, breaks, miscoven, yarn breaks, etc., which can lead to a decrease in quality of the fabric and even affect the performance and appearance of the final product. Therefore, developing a method that can quickly identify and quantitatively evaluate mesh defects is critical to quality control of the manufacturing process.
Conventional detection methods generally rely on manual visual inspection or simple mechanical equipment, and although the methods can be used for inspecting large batches of mesh fabrics to be shipped, manual inspection is time-consuming and labor-consuming, subjectivity and error can exist, defective mesh fabrics are omitted, and false alarm or missing alarm problems occur, which can affect the large-scale popularization of the mesh fabrics in the market.
Disclosure of Invention
(one) solving the technical problems
Aiming at the defects of the prior art, the invention provides a mesh fabric defect rapid detection method based on image data processing, which adopts image acquisition equipment and convolutional neural network technology, can efficiently and rapidly detect the defects of the surface of the mesh fabric, and does not need to consume a great deal of manpower and time; the quality and the analyzability of the image are effectively improved through multiple technologies such as preprocessing, denoising, smoothing, geometric correction and the like, and the defect can be further accurately identified and extracted; the image frame is uniformly divided into a plurality of netlike detection areas, so that each area can be used for detecting and evaluating defects independently, and the parallelism and efficiency of detection are improved; through deep calculation and learning, comprehensively considering evaluation indexes of a plurality of factors of the defect area, and finally obtaining a distortion density indexBy setting the preset distortion threshold K and the threshold Q, the method can carry out custom setting according to specific production requirements and quality standards so as to be suitable for different types and grades of mesh fabrics and locate defective areas and non-defective areas in image framesA domain from which a defect level report for the mesh fabric is obtained; in a word, compared with the prior art, the method further reduces labor cost and defective rate, and further improves the quality of the net fabric by accurately detecting and evaluating defects, so that the final product meets the quality requirement of high standards.
(II) technical scheme
In order to achieve the above purpose, the invention is realized by the following technical scheme: a mesh fabric defect rapid detection method based on image data processing comprises the following steps,
shooting and acquiring continuous electronic images of the mesh fabric to be detected by using image acquisition equipment in advance, and acquiring continuous image frames related to recording;
preprocessing continuous image frames on the surface of the mesh fabric, removing noise, smoothing images and performing geometric correction, identifying and extracting effective features in the continuous image frames, and generating an image dataset according to the image enhancement technology to improve the quality of the images, so that the subsequent analysis is facilitated;
uniformly dividing an image frame to be detected into a plurality of net-shaped detection areas, setting detection points in the net-shaped detection areas, detecting defect states, determining the defect positions of the surface of the net-shaped fabric, and marking the defect positions with the defect center positions as marks to serve as defect referencesBy defect reference->The positions are diffused to the periphery, and the convolutional neural network learning technology is utilized to detect and obtain the production area of the surface defects of the mesh fabric to be detected>And compare it with the sample area of a standard mesh surfaceGenerating defect proportion value by comparison>;
According to the defect proportion valueComparing and analyzing with a preset threshold value Q to determine a plurality of net-shaped detection areas as defect areas and non-defect areas;
acquiring centrifugal spacing of mesh fabric defects in defective areas and non-defective areas sequentiallyDifference in angle->Fabric thickness->And learning the data through deep calculation to obtain: distortion Density index->After dimensionless treatment, the distortion density index +.>Obtained by the following formula:
wherein->Expressed as a normalization function>Expressed as the number of defective areas>Denoted as->Distortion factor in the individual defect area>And (3) summing;
using distortion density indexAs a result, it is compared with a preset distortion threshold K for rapid detection of the defect level report of the mesh fabric.
Preferably, the image acquisition device is used for shooting and acquiring continuous electronic images of the mesh fabric to be detected and continuously recording related image frames so as to ensure that the whole mesh fabric is completely recorded, and various mesh fabrics in the images and edge, corner, color saturation, color splicing edge, shape and texture data information of the mesh fabric are identified, and the continuous image frames are sequentially ordered according to the data information and shooting sequence in the images and the corresponding sequence so as to detect a plurality of defect reference in the continuous image framesAdjacent spaces.
Preferably, the wavelet denoising technology is utilized, the data information in a plurality of image frames is decomposed through wavelet transformation, noise components are removed in a wavelet domain, the plurality of image frames are reconstructed back to an original domain through inverse transformation, histogram equalization is adopted to adjust brightness, the image range is stretched, and the contrast of an image is adjusted;
the key characteristic information in the image frame is extracted through the image enhancement technology, the defect part in the image is identified and analyzed, and the difference obvious degree between different areas in the image frame is selected and adjusted according to the characteristics of the defect part.
Preferably, algorithm model training and analysis are performed by using convolutional neural network learning technology to obtain: defect ratio valueOffset coefficient->Covering influence coefficient->And the offset coefficient ∈ ->And the coverage influence coefficient->Associated, obtain distortion factor->。
Preferably, the continuous image frame of the surface of the mesh fabric is divided into a plurality of mesh detection areas with equal area in the form of a grid, and a plurality of detection points are arranged in each mesh detection area to rapidly detect the defect state of the mesh fabric so as to search the defect position of the surface of the mesh fabric and obtain the defect proportion value in one of the mesh detection areasAfter dimensionless treatment, the defect proportion value +.>Obtained by the following formula:
wherein->Expressed as defect production area>Expressed as specimen area;
analyzing and obtaining defect proportion value in continuous image frames of surface of mesh fabricIf the defect proportion value->When the detection area is larger than a preset threshold value Q, indicating that the mesh detection area contains a defect area, and marking a serial number for the defect area; if the defect proportion value->When the detection area is smaller than a preset threshold value Q, the detection area does not contain a defect area, is a non-defect area, and is marked with a serial number.
Preferably, the centrifugal spacing is setIs +.>Correlating and obtaining the offset coefficient after dimensionless technical treatment>The offset coefficient->Obtained by the following formula:
wherein->Expressed as textile density>Andcentrifugal distance->Difference in angle->And textile Density->Weight value of (2);
wherein,,/>,/>and (2) and,/>expressed as a constant correction coefficient.
Preferably, the defect proportion valueAnd transparency->Correlating and obtaining the coverage influence coefficient after dimensionless technical treatment>The coverage influence coefficient +.>Obtained by the following formula:
wherein->Expressed as fabric thickness>And->Are all denoted as weight values +.>Expressed as a constant correction coefficient;
wherein,,/>and->。
Preferably, the coverage influence coefficient is combinedAnd offset coefficient->As a result, depth mining calculation is performed to obtain distortion factor +.>After dimensionless calculation processing, the distortion factor +.>Obtained by the following formula:
wherein->Expressed as an overlay influence coefficient->Is used for the weight value of (a),expressed as offset coefficient->Weight value of->Expressed as a constant correction factor, wherein +.>,/>And->。
Preferably, according to the distortion factor in a defective regionObtaining distortion Density index in all defective areas>And distortion Density index->And (3) comparing and analyzing with a preset distortion threshold K to obtain a defect grade report of the mesh fabric, and adopting a corresponding reworking scheme in a targeted manner.
Preferably, the preset distortion threshold K includes a first preset threshold K1 and a second preset threshold K2, where the first preset threshold K1 is greater than the preset distortion threshold K, and the preset distortion threshold K is greater than the second preset threshold K2;
if the second preset threshold value K2 is less than or equal to the distortion density indexWhen the preset distortion threshold value K is less than or equal to the preset distortion threshold value K, the current detection of the yarn or knitting wool treatment on the surface of the mesh fabric is not abnormal, the detection is qualified, and reworking treatment is not needed;
if the preset distortion threshold K is less than or equal to the distortion density indexWhen the first preset threshold value K1 is less than or equal to the first preset threshold value K1, the current detection of partial flaw of the surface yarn or knitting wool treatment of the mesh fabric is shown, the detection is unqualified, and the defect reference is determined>And reworking;
if the first preset threshold value K1 is less than or equal to the distortion density indexWhen the current detection of the surface yarn or knitting wool treatment of the mesh fabric has large area defects, the detection standard is seriously failed, and the defect standard is determined>And (5) immediately reworking.
(III) beneficial effects
The invention provides a mesh fabric defect rapid detection method based on image data processing. The beneficial effects are as follows:
(1) The method for rapidly detecting the defects of the mesh fabric based on image data processing adopts image acquisition equipment and convolutional neural network technology, so that the defects on the surface of the mesh fabric can be efficiently and rapidly detected, and a large amount of manpower and time are not required to be consumed; the quality and the analyzability of the image are effectively improved through multiple technologies such as preprocessing, denoising, smoothing, geometric correction and the like, and the defect can be further accurately identified and extracted; the image frame is uniformly divided into a plurality of netlike detection areas, so that each area can be used for detecting and evaluating defects independently, and the parallelism and efficiency of detection are improved; through deep calculation and learning, comprehensively considering evaluation indexes of a plurality of factors of the defect area, and finally obtaining a distortion density indexBy setting the preset distortion threshold K and the preset distortion threshold Q, the self-defining setting can be carried out according to specific production requirements and quality standards so as to be suitable for different types and grades of mesh fabrics, and defect areas and non-defect areas in image frames are positioned and defect grade reports of the mesh fabrics are obtained; in summary, compared with the prior art, the method further reduces labor cost and defective rate, and further improves by accurately detecting and evaluating defectsThe quality of the high-mesh fabric ensures that the final product meets the quality requirements of high standards.
(2) According to the mesh fabric defect rapid detection method based on image data processing, continuous electronic image frames of a mesh fabric are comprehensively shot and continuously recorded by using image acquisition equipment, so that all parts of the whole fabric are ensured to be completely recorded, comprehensive detection information is provided for further accurately analyzing and detecting defect conditions, and the positions and distribution rules of defects are facilitated to be detected by sequencing the continuous image frames according to the image shooting sequence; noise in the image is removed through a wavelet denoising technology and an image enhancement technology, so that the image is clearer, defects can be accurately detected, and defective areas and non-defective areas can be further accurately distinguished by adjusting the obvious degree of difference between different areas.
(3) According to the mesh fabric defect rapid detection method based on image data processing, the distortion condition of the surface of the mesh fabric is comprehensively evaluated through the data influence of multi-dimensional factors, and the distortion factor is obtainedThe comprehensive evaluation more comprehensively reflects the quality and defect condition of the fabric; at the same time, the distortion Density index->Comparing with a preset distortion threshold K, providing a multi-situation defect grade report so as to determine the severity of the defect, and taking corresponding reworking measures; according to different distortion degrees, a reworking measure is adopted in a targeted manner, and the fabric which is seriously unqualified can be reworked immediately so as to reduce the defective rate.
Drawings
FIG. 1 is a block diagram and schematic flow chart of a method for rapidly detecting defects of a mesh fabric based on image data processing according to the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In the field of mesh fabrics based on image data processing, the production and quality control of mesh fabrics requires efficient and accurate methods to detect and evaluate defects thereof, including problems of holes, breaks, miscoven, yarn breaks, etc., which can lead to a decrease in quality of the fabric and even affect the performance and appearance of the final product. Therefore, developing a method that can quickly identify and quantitatively evaluate mesh defects is critical to quality control of the manufacturing process.
Conventional detection methods generally rely on manual visual inspection or simple mechanical equipment, and although the methods can be used for inspecting large batches of mesh fabrics to be shipped, manual inspection is time-consuming and labor-consuming, subjectivity and error can exist, defective mesh fabrics are omitted, and false alarm or missing alarm problems occur, which can affect the large-scale popularization of the mesh fabrics in the market.
Example 1
Referring to fig. 1, the present invention provides a method for rapidly detecting defects of a mesh fabric based on image data processing, which comprises the following steps,
shooting and acquiring continuous electronic images of the mesh fabric to be detected by using image acquisition equipment in advance, and acquiring continuous image frames related to recording;
preprocessing continuous image frames on the surface of the mesh fabric, removing noise, smoothing images and performing geometric correction, identifying and extracting effective features in the continuous image frames, and generating an image dataset according to the image enhancement technology to improve the quality of the images, so that the subsequent analysis is facilitated;
uniformly dividing an image frame to be detected into a plurality of net-shaped detection areas, setting detection points in the net-shaped detection areas, detecting defect states, determining the defect positions of the surface of the net-shaped fabric, marking the defect positions with the defect center positions as marks, and marking the positions asDefect referenceBy defect reference->The positions are diffused to the periphery, and the convolutional neural network learning technology is utilized to detect and obtain the production area of the surface defects of the mesh fabric to be detected>And compare it with the sample area of a standard mesh surfaceGenerating defect proportion value by comparison>;
According to the defect proportion valueComparing and analyzing with a preset threshold value Q to determine a plurality of net-shaped detection areas as defect areas and non-defect areas;
acquiring centrifugal spacing of mesh fabric defects in defective areas and non-defective areas sequentiallyDifference in angle->Fabric thickness->And learning the data through deep calculation to obtain: distortion Density index->After dimensionless treatment, distortion Density index +.>Obtained by the following formula:
wherein->Expressed as a normalization function>Expressed as the number of defective areas>Denoted as->Distortion factor in the individual defect area>And (3) summing;
using distortion density indexAs a result, it is compared with a preset distortion threshold K for rapid detection of the defect level report of the mesh fabric.
In the method, the defects on the surface of the mesh fabric can be efficiently and rapidly detected by using image acquisition equipment and convolutional neural network technology; the image frame is uniformly divided into a plurality of netlike detection areas, so that each area can be independently subjected to defect detection and evaluation; through deep calculation and learning, comprehensively considering evaluation indexes of a plurality of factors of the defect area, and finally obtaining a distortion density indexBy setting the preset distortion threshold K and the preset distortion threshold Q, the preset distortion threshold K and the preset distortion threshold Q can be customized according to specific production requirements and quality standards so as to be suitable for different types and grades of mesh fabrics, and defective areas and non-defective areas in image frames can be positioned and defect grade reports of the mesh fabrics can be obtained.
Example 2
Referring to fig. 1, the following details are: using image acquisition equipment to shoot and acquire continuous electronic images of the mesh fabric to be detected and continuously record related image frames so as to ensure that the whole mesh fabric is completely recorded, identifying various mesh fabrics in the images and edges, corner points, color saturation, color splicing edges, shapes and texture data information of the mesh fabric, and sequentially sequencing the continuous image frames according to the data information and shooting sequence in the images and the corresponding sequence so as to detect a plurality of defect references in the continuous image framesAdjacent spaces.
Decomposing data information in a plurality of image frames by wavelet transformation by utilizing a wavelet denoising technology, removing noise components in a wavelet domain, reconstructing the plurality of image frames back to an original domain by inverse transformation, adjusting brightness by adopting histogram equalization, stretching an image range, and adjusting contrast of an image;
the key characteristic information in the image frame is extracted through the image enhancement technology, the defect part in the image is identified and analyzed, and the difference obvious degree between different areas in the image frame is selected and adjusted according to the characteristics of the defect part.
In the embodiment, continuous electronic image frames of the mesh fabric are comprehensively shot and continuously recorded by using the image acquisition equipment, so that all parts of the whole fabric are ensured to be completely recorded, and therefore, comprehensive detection information is provided for further accurately analyzing and detecting the defect condition, and the position and distribution rule of defects are facilitated to be detected by sequencing the continuous image frames according to the image shooting sequence; noise in the image is removed through a wavelet denoising technology and an image enhancement technology, so that the image is clearer, defects can be accurately detected, and defective areas and non-defective areas can be further accurately distinguished by adjusting the obvious degree of difference between different areas.
Example 3
Referring to fig. 1, the following details are: and (3) training and analyzing an algorithm model by utilizing a convolutional neural network learning technology to obtain: lack of supplyTrap ratio valueOffset coefficient->Covering influence coefficient->And the offset coefficient ∈ ->And the coverage influence coefficient->Associated, obtain distortion factor->。
Dividing continuous image frames of the surface of the mesh fabric into a plurality of mesh detection areas with equal areas in a grid form, setting a plurality of detection points in each mesh detection area to rapidly detect the defect state of the mesh fabric so as to search the defect position of the surface of the mesh fabric and obtain the defect proportion value in one mesh detection areaDefect ratio value +.>Obtained by the following formula:
wherein->Expressed as defect production area>Expressed as specimen area;
defect production areaRefers to various defect areas on the surface of the mesh fabric during the production process, wherein defects comprise holes, embroidery coverage, chromatic aberration, yarn dislocation or kinking and the like;
specimen areaRefers to a standard mesh fabric area;
analyzing and obtaining defect proportion value in continuous image frames of surface of mesh fabricIf the defect proportion value->When the detection value is larger than a preset threshold value Q, indicating that the mesh detection area contains a defect area, and marking a serial number for the defect area; if the defect proportion value->When the detection area is smaller than a preset threshold value Q, the detection area does not contain a defect area, the detection area is a non-defect area, and serial numbers are marked on the non-defect area.
Will centrifuge the distanceDifference from angle->Correlating and obtaining the offset coefficient after dimensionless technical treatment>Offset coefficient->Obtained by the following formula:
wherein->Expressed as textile density>Andcentrifugal distance->Difference in angle->And textile Density->Weight value of (2);
wherein,,/>,/>and (2) and,/>expressed as a constant correction coefficient.
The above-mentioned textile densityThe degree of tightness of spinning is referred, and the spinning degree is acquired through a spinning counter;
centrifugal spacingRefers to the distance of the defective area from the original position, the angle difference +.>Refers to the defect area being offset from the original position by an angle, and the centrifugal spacing +.>And angle difference->All are acquired by computer vision technology.
Defect ratio valueAnd transparency->Correlating and obtaining the coverage influence coefficient after dimensionless technical treatment>Covering influence coefficient->Obtained by the following formula:
wherein->Expressed as fabric thickness, including thickness of local embroidery and thickness of decorations, < >>And->Are all denoted as weight values +.>Expressed as a constant correction coefficient;
wherein,,/>and->。
Transparency of the filmTransparency including mesh fabric surface coverings;
thickness of the fabricAnd transparency->All are acquired by computer vision technology.
In the present embodiment, the defect ratio value is obtained byThe defect area and the non-defect area in the mesh fabric surface image frame can be judged, and the technology such as convolutional neural network is used, so that automatic defect detection is realized, the detection speed and efficiency are further improved, and the artificial subjective error is reduced; the data information of the mesh fabrics with multiple dimensions is collected, so that the defect severity of the mesh fabrics can be evaluated more comprehensively, and a powerful foundation is provided for improving the quality of the fabrics subsequently.
Example 4
Referring to fig. 1, the following details are: combined coverage influence coefficientAnd offset coefficient->As a result, depth mining calculation is performed to obtain distortion factor +.>After dimensionless calculation processing, distortionFactor->Obtained by the following formula:
wherein->Expressed as an overlay influence coefficient->Is used for the weight value of (a),expressed as offset coefficient->Weight value of->Expressed as a constant correction factor, wherein +.>,/>And->。
According to distortion factors in a defective regionObtaining distortion Density index in all defective areas>And distortion Density index->Comparing and analyzing with a preset distortion threshold K to obtain a defect grade report of the mesh fabric, and adopting corresponding modes in a targeted mannerReworking scheme.
The preset distortion threshold K comprises a first preset threshold K1 and a second preset threshold K2, wherein the first preset threshold K1 is larger than the preset distortion threshold K, and the preset distortion threshold K is larger than the second preset threshold K2;
if the second preset threshold value K2 is less than or equal to the distortion density indexWhen the preset distortion threshold value K is less than or equal to the preset distortion threshold value K, the current detection of the yarn or knitting wool treatment on the surface of the mesh fabric is not abnormal, the detection is qualified, and reworking treatment is not needed;
if the preset distortion threshold K is less than or equal to the distortion density indexWhen the first preset threshold value K1 is less than or equal to the first preset threshold value K1, the current detection of partial flaw of the surface yarn or knitting wool treatment of the mesh fabric is shown, the detection is unqualified, and the defect reference is determined>And reworking;
if the first preset threshold value K1 is less than or equal to the distortion density indexWhen the current detection of the surface yarn or knitting wool treatment of the mesh fabric has large area defects, the detection standard is seriously failed, and the defect standard is determined>And (5) immediately reworking.
In the present embodiment, by taking into consideration the coverage influence coefficientAnd offset coefficient->Two factors are given different weights, the distortion condition of the surface of the mesh fabric is comprehensively evaluated, and a distortion factor is obtained>The comprehensive evaluation more comprehensively reflects the quality and defect condition of the fabric; at the same time, the distortion Density index->Comparing with a preset distortion threshold K, providing a multi-situation defect grade report so as to determine the severity of the defect, and taking corresponding reworking measures; according to different distortion degrees, a reworking measure is adopted in a targeted manner, and the fabric which is seriously unqualified can be reworked immediately so as to reduce the defective rate.
Examples: a certain factory for manufacturing mesh fabric, which introduces a mesh fabric defect rapid detection method based on image data processing, is the following example of the factory for manufacturing mesh fabric:
image data acquisition: defect production area5; specimen area->23; centrifugal distance->3.21; angle difference->6.5 degrees; textile Density->76%; fabric thickness->2.2; transparency->84%; />0.47; />0.32; />0.63; />0.42; />0.55; />0.42; />0.59; />Is 2; />4; />7;
from the above data, the following calculations can be made:
defect ratio value=/>=0.22;
Coefficient of offset=/>=4.54+0.49+2.0=7.03;
Coefficient of influence of coverage=/>=16.23;
Distortion factor=/>=6.82+4.15+7=17.97;
Distortion density index=/>=202.61;
If the preset distortion threshold K is 200, the preset distortion threshold K is less than or equal to the distortion density index when the first preset threshold K1 is 210 and the second preset threshold K2 is 180A first preset threshold value K1 which is less than or equal to the first preset threshold value, and is expressed as that the yarn or knitting wool treatment on the surface of the mesh fabric which is currently detected has partial flaws, the detection is unqualified, and a defect reference is required to be determined>And reworking.
Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (7)
1. A mesh fabric defect rapid detection method based on image data processing is characterized in that: including the following that are included in the description,
shooting and acquiring continuous electronic images of the mesh fabric to be detected by using image acquisition equipment in advance, and acquiring continuous image frames related to recording;
preprocessing continuous image frames on the surface of the mesh fabric, removing noise, smoothing images and performing geometric correction, identifying and extracting effective features in the continuous image frames, and generating an image dataset according to the image enhancement technology to improve the quality of the images, so that the subsequent analysis is facilitated;
uniformly dividing an image frame to be detected into a plurality of net-shaped detection areas, setting detection points in the net-shaped detection areas, detecting defect states, determining the defect positions of the surface of the net-shaped fabric, and marking the defect positions with the defect center positions as marks to serve as defect referencesBy defect reference->The positions are diffused to the periphery, and the convolutional neural network learning technology is utilized to detect and obtain the production area of the surface defects of the mesh fabric to be detected>And compare it with the sample area of a standard mesh surfaceGenerating defect proportion value by comparison>;
According to the defect proportion valueComparing and analyzing with a preset threshold value Q to determine a plurality of net-shaped detection areas as defect areas and non-defect areas;
acquiring centrifugal spacing of mesh fabric defects in defective areas and non-defective areas sequentiallyDifference in angle->Fabric thickness->Textile Density->The degree of tightness of spinning is referred, and the spinning degree is acquired through a spinning counter; the centrifugal distance->Refers to the distance of the defective area from the original position, the angle difference +.>Refers to the defect area being offset from the original position by an angle, and the centrifugal spacing +.>And angle difference->All are acquired by a computer vision technology;
the centrifugal spacing is setIs +.>Correlating and obtaining the offset coefficient after dimensionless technical treatment>The offset coefficient->Obtained by the following formula:
in (1) the->Expressed as textile density>And->Centrifugal distance->Difference in angle->And textile Density->Weight value of (2);
wherein,,/>,/>and->,/>Expressed as a constant correction coefficient;
acquisition of transparency by computer vision techniquesTransparency->Transparency including mesh fabric surface coverings;
defect ratio valueAnd transparency->Correlating and obtaining coverage influence coefficients after processing by dimensionless technologyThe coverage influence coefficient +.>Obtained by the following formula:
in (1) the->Expressed as fabric thickness>And->Are all denoted as weight values +.>Expressed as a constant correction coefficient;
wherein,,/>and->;
Combined coverage influence coefficientAnd offset coefficient->As a result, depth mining calculation is performed to obtain distortion factorsAfter dimensionless calculation processing, the distortion factor +.>Obtained by the following formula:
in (1) the->Expressed as an overlay influence coefficient->Weight value of->Expressed as offset coefficient->Weight value of->Expressed as a constant correction factor, wherein +.>,/>And (2) and;
and the method comprises the steps of obtaining by deep calculation learning: distortion density indexAfter dimensionless treatment, the distortion density index +.>Obtained by the following formula:
in (1) the->Expressed as a normalization function>Expressed as the number of defective areas>Denoted as->Distortion factor in the individual defect area>And (3) summing;
using distortion density indexAs a result, it is compared with a preset distortion threshold K for rapid detection of the defect level report of the mesh fabric.
2. The method for rapidly detecting defects in a mesh fabric based on image data processing according to claim 1, wherein: using image-capturing devicesCapturing continuous electronic images of the mesh fabric to be detected, and continuously recording related image frames to ensure complete recording of the whole mesh fabric, identifying various mesh fabrics in the images and edges, corner points, color saturation, color splicing edges, shapes and texture data information of the mesh fabric, and sequentially sequencing the continuous image frames according to the data information and the capturing sequence in the images according to the corresponding sequence so as to detect a plurality of defect references in the continuous image framesAdjacent spaces.
3. The method for rapidly detecting defects in a mesh fabric based on image data processing according to claim 2, wherein: decomposing data information in a plurality of image frames by wavelet transformation by utilizing a wavelet denoising technology, removing noise components in a wavelet domain, reconstructing the plurality of image frames back to an original domain by inverse transformation, adjusting brightness by adopting histogram equalization, stretching an image range, and adjusting contrast of an image;
the key characteristic information in the image frame is extracted through the image enhancement technology, the defect part in the image is identified and analyzed, and the difference obvious degree between different areas in the image frame is selected and adjusted according to the characteristics of the defect part.
4. A method for rapid detection of mesh defects based on image data processing as claimed in claim 3, wherein: and (3) training and analyzing an algorithm model by utilizing a convolutional neural network learning technology to obtain: defect ratio valueOffset coefficient->Covering influence coefficient->And the offset coefficient ∈ ->And the coverage influence coefficient->Associated, obtain distortion factor->。
5. The method for rapidly detecting defects in a mesh fabric based on image data processing according to claim 4, wherein: dividing continuous image frames of the surface of the mesh fabric into a plurality of mesh detection areas with equal areas in a grid form, setting a plurality of detection points in each mesh detection area to rapidly detect the defect state of the mesh fabric so as to search the defect position of the surface of the mesh fabric and obtain the defect proportion value in one mesh detection areaAfter dimensionless treatment, the defect proportion value +.>Obtained by the following formula:
in (1) the->Expressed as defect production area>Expressed as specimen area;
analyzing and obtaining defect proportion value in continuous image frames of surface of mesh fabricIf the defect proportion value->When the detection area is larger than a preset threshold value Q, indicating that the mesh detection area contains a defect area, and marking a serial number for the defect area; if the defect proportion value->When the detection area is smaller than a preset threshold value Q, the detection area does not contain a defect area, is a non-defect area, and is marked with a serial number.
6. The method for rapidly detecting defects in a mesh fabric based on image data processing according to claim 5, wherein: according to distortion factors in a defective regionObtaining distortion Density index in all defective areas>And distortion Density index->And (3) comparing and analyzing with a preset distortion threshold K to obtain a defect grade report of the mesh fabric, and adopting a corresponding reworking scheme in a targeted manner.
7. The method for rapidly detecting defects in a mesh fabric based on image data processing according to claim 6, wherein: the preset distortion threshold K comprises a first preset threshold K1 and a second preset threshold K2, wherein the first preset threshold K1 is larger than the preset distortion threshold K, and the preset distortion threshold K is larger than the second preset threshold K2;
if the second preset threshold value K2 is less than or equal to the distortion density indexWhen the preset distortion threshold value K is less than or equal to the preset distortion threshold value K, the current detection of the yarn or knitting wool treatment on the surface of the mesh fabric is not abnormal, the detection is qualified, and reworking treatment is not needed;
if the preset distortion threshold K is less than or equal to the distortion density indexWhen the first preset threshold value K1 is less than or equal to the first preset threshold value K1, the current detection of partial flaw of the surface yarn or knitting wool treatment of the mesh fabric is shown, the detection is unqualified, and the defect reference is determined>And reworking;
if the first preset threshold value K1 is less than or equal to the distortion density indexWhen the current detection of the surface yarn or knitting wool treatment of the mesh fabric has large area defects, the detection standard is seriously failed, and the defect standard is determined>And (5) immediately reworking.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311462529.0A CN117218114B (en) | 2023-11-06 | 2023-11-06 | Mesh fabric defect rapid detection method based on image data processing |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311462529.0A CN117218114B (en) | 2023-11-06 | 2023-11-06 | Mesh fabric defect rapid detection method based on image data processing |
Publications (2)
Publication Number | Publication Date |
---|---|
CN117218114A CN117218114A (en) | 2023-12-12 |
CN117218114B true CN117218114B (en) | 2024-01-30 |
Family
ID=89049570
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311462529.0A Active CN117218114B (en) | 2023-11-06 | 2023-11-06 | Mesh fabric defect rapid detection method based on image data processing |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117218114B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117437235B (en) * | 2023-12-21 | 2024-03-12 | 四川新康意众申新材料有限公司 | Plastic film quality detection method based on image processing |
CN117876342A (en) * | 2024-01-15 | 2024-04-12 | 泰州市华亿源机械有限公司 | Adaptive gear defect detection method based on deep learning |
CN117974595A (en) * | 2024-01-30 | 2024-05-03 | 苏州符兹达电子科技有限公司 | Circuit board defect detection method and system applying image processing technology |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108076344A (en) * | 2017-11-27 | 2018-05-25 | 上海交通大学 | A kind of VR video code rate control bit distribution methods based on weight graph model |
CN111179229A (en) * | 2019-12-17 | 2020-05-19 | 中信重工机械股份有限公司 | Industrial CT defect detection method based on deep learning |
CN112630222A (en) * | 2020-11-24 | 2021-04-09 | 河海大学常州校区 | Mobile phone cover plate glass defect detection method based on machine vision |
CN114170208A (en) * | 2021-12-14 | 2022-03-11 | 武汉福旺家包装有限公司 | Paper product defect detection method based on artificial intelligence |
CN114842017A (en) * | 2022-07-06 | 2022-08-02 | 济宁市海富电子科技有限公司 | HDMI cable surface quality detection method and system |
CN114994073A (en) * | 2022-05-24 | 2022-09-02 | 上海华力集成电路制造有限公司 | Defect scanning detection analysis method of Die |
CN115575402A (en) * | 2022-10-17 | 2023-01-06 | 中国兵器装备集团西南技术工程研究所 | Intelligent identification, calculation and judgment method for defects of inner wall of closed cylindrical part |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE502007002821D1 (en) * | 2007-08-10 | 2010-03-25 | Sick Ag | Recording of equalized images of moving objects with uniform resolution by line sensor |
FR3034889B1 (en) * | 2015-04-10 | 2017-04-28 | Cn2P Sas | ELECTRONIC BRACELET FOR DISPLAYING AN INTERACTIVE DIGITAL CONTENT TO BE PROJECTED ON A ZONE OF AN ARM |
JP7042118B2 (en) * | 2018-03-08 | 2022-03-25 | 株式会社東芝 | Inspection equipment, inspection methods, and programs |
-
2023
- 2023-11-06 CN CN202311462529.0A patent/CN117218114B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108076344A (en) * | 2017-11-27 | 2018-05-25 | 上海交通大学 | A kind of VR video code rate control bit distribution methods based on weight graph model |
CN111179229A (en) * | 2019-12-17 | 2020-05-19 | 中信重工机械股份有限公司 | Industrial CT defect detection method based on deep learning |
CN112630222A (en) * | 2020-11-24 | 2021-04-09 | 河海大学常州校区 | Mobile phone cover plate glass defect detection method based on machine vision |
CN114170208A (en) * | 2021-12-14 | 2022-03-11 | 武汉福旺家包装有限公司 | Paper product defect detection method based on artificial intelligence |
CN114994073A (en) * | 2022-05-24 | 2022-09-02 | 上海华力集成电路制造有限公司 | Defect scanning detection analysis method of Die |
CN114842017A (en) * | 2022-07-06 | 2022-08-02 | 济宁市海富电子科技有限公司 | HDMI cable surface quality detection method and system |
CN115575402A (en) * | 2022-10-17 | 2023-01-06 | 中国兵器装备集团西南技术工程研究所 | Intelligent identification, calculation and judgment method for defects of inner wall of closed cylindrical part |
Non-Patent Citations (2)
Title |
---|
A single-phase synchronization technique for grid-connected RESS under distorted grid conditions;Komal Saleem 等;《2019 IEEE Energy Conversion Congress and Exposition (ECCE)》;第4665-4672页 * |
带钢隆起缺陷成因及控制措施的研究;夏银锋 等;《轧钢》;第35卷(第5期);第75-78页 * |
Also Published As
Publication number | Publication date |
---|---|
CN117218114A (en) | 2023-12-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN117218114B (en) | Mesh fabric defect rapid detection method based on image data processing | |
CN114549522B (en) | Textile quality detection method based on target detection | |
CN113989279B (en) | Plastic film quality detection method based on artificial intelligence and image processing | |
Liu et al. | Steel surface defect detection using a new Haar–Weibull-variance model in unsupervised manner | |
CN110349126B (en) | Convolutional neural network-based marked steel plate surface defect detection method | |
CN113554080A (en) | Non-woven fabric defect detection and classification method and system based on machine vision | |
CN115063409B (en) | Method and system for detecting surface material of mechanical cutter | |
CN113838054B (en) | Mechanical part surface damage detection method based on artificial intelligence | |
CN114627125B (en) | Stainless steel tablet press surface quality evaluation method based on optical means | |
CN116758491B (en) | Printing monitoring image analysis method and system applied to 3D printing | |
CN107966444B (en) | Textile flaw detection method based on template | |
CN108444934B (en) | Automatic segmentation and quantification method for debonding defect of composite material | |
CN114757900A (en) | Artificial intelligence-based textile defect type identification method | |
CN117273554B (en) | Textile production quality prediction method based on data identification | |
CN109239073A (en) | A kind of detection method of surface flaw for body of a motor car | |
CN116152242B (en) | Visual detection system of natural leather defect for basketball | |
CN117152119A (en) | Profile flaw visual detection method based on image processing | |
CN117611583A (en) | Artificial intelligence-based aluminum composite panel defect detection method and system | |
CN115205209A (en) | Monochrome cloth flaw detection method based on weak supervised learning | |
CN115983687A (en) | Intelligent detection management system and method for quality of cold-rolled strip steel | |
Jing et al. | Detection of bobbin yarn surface defects by visual saliency analysis | |
CN116934752B (en) | Glass detection method and system based on artificial intelligence | |
CN113298857A (en) | Bearing defect detection method based on neural network fusion strategy | |
CN109785313B (en) | Tire qualification detection method based on LBP | |
CN114693652B (en) | Fabric Defect Detection Method Based on Gaussian Mixture Model |
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 |