CN114937013A

CN114937013A - Photovoltaic printing screen defect detection method and system

Info

Publication number: CN114937013A
Application number: CN202210545208.6A
Authority: CN
Inventors: 于龙; 雷力
Original assignee: Wuhan Guangmu Technology Co ltd
Current assignee: Wuhan Guangmu Technology Co ltd
Priority date: 2022-05-19
Filing date: 2022-05-19
Publication date: 2022-08-23

Abstract

The invention provides a method and a system for detecting defects of a photovoltaic printing screen, which comprise the following steps: determining pictures of the photovoltaic printing screen under the conditions of an upper light source and a lower light source; processing the collected lower light source picture, and measuring partial dynamic parameters of the screen printing plate; performing Blob analysis on the lower light source image, extracting a continuous region with the highest gray value in the lower light source image, and taking the continuous region as a grid line of the photovoltaic printing screen; removing the grid lines in the lower light source picture according to the extracted grid lines, and detecting the highlight area to detect the pinhole defect; performing Blob analysis on the grid lines in the extracted grid line positioning image to detect the grid blocking defect; removing the grid lines in the image according to the extracted grid lines, and then carrying out Blob division to carry out dirt taking defect detection; and detecting the edges of the steel wires in the lower light source image, and judging the positions of the edges to detect the broken wire defects. The invention can well solve the problems of low human eye detection efficiency and low Gerber dragging detection speed required by machine detection.

Description

Photovoltaic printing screen defect detection method and system

Technical Field

The invention belongs to the field of defect detection of printing screens, and particularly relates to a method and a system for detecting defects of a photovoltaic printing screen.

Background

With the development of society, photovoltaic has been developed vigorously by means of its characteristics of cleanness, no pollution, safety, reliability, convenience for maintenance and the like, and becomes an important direction for the development of new energy. The batch manufacturing of the photovoltaic cells mostly adopts a photovoltaic printing screen as a mould for printing. Therefore, the quality of the printing screen plate is highly dependent on the production process of the photovoltaic cell panel. Common defects on the surface of the screen printing plate include screen blocking, broken silk, dirt, pinholes and the like. These defects can affect the printing quality of the photovoltaic panel, and thus affect the normal use of the photovoltaic panel. It is therefore necessary to perform surface defect detection before the photovoltaic printing screen is put into photovoltaic cell production.

The surface defects of the common photovoltaic screen printing plate are detected by pinholes, screen blockage, dirt, thin glue and broken filaments. Wherein the pinhole is a bright spot defect appearing in the latex area of the screen printing plate and represents that the film of the screen printing plate is cracked or perforated; the blocking screen is used for blocking the defect of the grid line area of the screen printing plate; the dirt is the defect that latex is blocked and the shape is larger; the thin glue is transparent glue and covers the silk screen in the screen printing plate, and the defect can not be detected in a common backlight light source; the broken wire is the defect that the steel wire forming the screen plate is broken to destroy the grid structure.

At present, the defect detection of the photovoltaic cell screen printing plate still depends on manual detection, the human eye detection cost is high, the detection accuracy is limited, and the human eye is easy to fatigue and is not beneficial to long-time work.

In accordance with the great trend of artificial intelligence, the surface defect detection of the photovoltaic cell screen is performed by combining an industrial camera and a computer program by adopting a machine vision technology. The defects of manual detection are overcome to a certain extent, but the existing machine vision detection method detects the surface defects on the premise of reading a Gerber file for positioning grid lines when a printing screen is designed, the method is limited by the accuracy of a positioning algorithm, generally, the Gerber file needs to be cut and matched after being read in a large scale, and the accuracy and the detection speed of a subsequent defect detection algorithm are limited to a certain extent.

Therefore, the existing machine vision detection method needs to be improved, so that the accuracy and the working efficiency of the defect detection of the photovoltaic cell screen are improved, and the production quality of the photovoltaic printing screen and the cell is ensured.

Disclosure of Invention

Aiming at the defects of the prior art, the invention aims to provide a method and a system for detecting the defects of a photovoltaic printing screen plate, and aims to solve the problems of low accuracy and low efficiency of the defect detection of the photovoltaic printing screen plate.

In order to achieve the above object, in a first aspect, the present invention provides a method for detecting defects of a photovoltaic printing screen, including the following steps:

determining images of the photovoltaic printing screen under the condition of two light sources; the two light sources are: the upper light source is positioned right above the photovoltaic printing screen and the lower light source is positioned right below the photovoltaic printing screen; the image under the upper light source is a reflection type illumination image, and the image under the lower light source is a transmission type illumination image; the photovoltaic printing screen plate may have defects, and the defects include: pinholes, blocked nets, dirt and broken filaments; the blocking net is arranged in the grid line area; the steel wire of the broken wire finger printing screen plate is broken;

measuring dynamic parameters of the photovoltaic printing screen plate based on the lower light source image, wherein the dynamic parameters comprise: mesh width, steel wire width and steel wire inclination; the mesh width refers to the pixel width of a single grid in the photovoltaic printing screen;

performing Blob analysis on the lower light source image, extracting a continuous region with the highest gray value in the lower light source image, and taking the continuous region as a grid line of the photovoltaic printing screen;

removing grid lines in the lower light source image based on the extracted grid lines, and analyzing highlight area information on the lower light source image after the grid lines are removed to perform pinhole detection; when the area of a certain highlight area exceeds a preset area, the area is considered to be a pinhole correspondingly;

cutting a lower light source image by using the extracted grid lines to generate a first gray-scale image only containing the grid lines, analyzing and extracting a blocking net in the first gray-scale image only containing the grid lines by using Blob, subtracting two gray-scale images corresponding to an upper light source image and a lower light source image to obtain a subtracted gray-scale image, cutting the subtracted gray-scale image by using the extracted grid lines to generate a second gray-scale image only containing the grid lines, performing Blob analysis on the second gray-scale image only containing the grid lines, and further extracting the blocking net which is possibly missed to be detected in the two images; the gray scale image only containing the grid lines is a gray scale image with the pixel value of 0 in other areas of the image except the grid line area;

removing grid lines in the lower light source image and the upper light source image based on the extracted grid lines, and performing Blob analysis on the two images with the grid lines removed based on the mesh width to detect the dirt of the screen printing plate;

and removing the grid lines in the lower light source image based on the extracted grid lines, detecting the steel wires in the lower light source image after the grid lines are removed according to the width and the gradient of the steel wires, and performing broken wire detection on the edges of the steel wires by utilizing Blob analysis.

Optionally, the lower light source image and the upper light source image are both grayscale images.

In an optional example, extracting a blocking net in an image by Blob analysis specifically includes:

extracting the area of which the gray value is smaller than a first preset value in the image;

if the height of the extraction area is larger than the preset pixel and the area is larger than the mesh width, the area is considered as a blocking net at the joint of the grid line and the steel wire;

if the area of the extraction region is larger than the mesh width and the radius of the inscribed circle is larger than the preset radius, the region is considered as a blocked net in the grid line;

if the width of the extraction region is between the preset widths, the height is between the first preset heights, and the compactness is smaller than the threshold value, the region is considered as the grid blocking at the edge of the grid line.

In an optional example, performing Blob analysis on the image to detect contamination of the halftone, specifically:

after grid lines in an image are determined, setting regions in the image, which are less than a preset distance away from the grid lines, as near grid regions and setting regions in which the distance is greater than the preset distance away from the grid lines as far grid regions according to position information of the grid lines in the image;

removing steel wires in the near-grid region according to the width of the steel wires and the gradient of the steel wires to obtain a first steel wire removing image, and segmenting the first steel wire removing image by using a gray threshold to extract a first suspected dirty region with a gray value smaller than a second preset value;

if the height of the first suspected dirty area is between the second preset height and the area is larger than the mesh width, taking the first suspected dirty area as the dirty area of the near-grid area;

removing steel wires in a far grid region according to the width of the steel wires and the gradient of the steel wires to obtain a second steel wire removing map, and segmenting and extracting a second suspected dirty region with the gray value smaller than a third preset value from the second steel wire removing map by using a gray threshold value; the third preset value is a gray value corresponding to the trough after gray histogram statistics is carried out on the second type of steel wire removing image;

if the area of the second suspected smudged area is greater than twice the mesh width, it is considered as smudged in the far gate area.

In an optional example, the Blob analysis is used for performing the wire breakage detection on the edge of the steel wire, specifically:

if the edge of the steel wire is positioned in the image, the edge area is considered to be broken filaments; the image inner finger is away from the area with preset pixels on the wide side and the narrow side of the image.

In an optional example, Blob analysis is performed on the lower light source image, a continuous region with the highest gray value in the lower light source image is extracted, and the continuous region is used as a grid line of the photovoltaic printing screen, specifically:

carrying out gray value closing operation on the lower light source image by using the rectangular structural elements, and then carrying out gray threshold segmentation to extract an area with a gray value larger than a preset value to obtain a first coarse grid area; performing gray value closing operation on the lower light source image by using the hexagonal structural elements, and then performing gray threshold segmentation to extract a region with a gray value larger than a preset value to obtain a second coarse grid region; performing gray threshold segmentation on the lower light source image to obtain a third coarse grid region; wherein, the coarse grid region refers to a rough grid line region;

sequentially carrying out gray value closing operation, gray threshold segmentation, closing operation and opening operation on the first coarse grid region to obtain a vertical grid line;

carrying out gray value difference on the vertical grid lines by utilizing a third coarse grid region to obtain wave-shaped grid lines preliminarily, and shaping the wave-shaped grid lines preliminarily obtained through framework extraction to obtain final wave-shaped grid lines;

carrying out gray value difference on the vertical grid line and the final wavy grid line by using the second coarse grid region to obtain a pattern grid line; the pattern grid lines refer to grid lines except for vertical grid lines and wave-shaped grid lines;

combining the vertical grid lines, the wave-shaped grid lines and the pattern grid lines to obtain combined grid line regions; and extracting the gray characteristic of the grid line region to obtain the grid line of the photovoltaic printing screen.

In an optional example, the gray value closing operation, the gray threshold segmentation, the closing operation, and the opening operation are performed on the first coarse gate region to obtain a vertical gate line, which specifically includes:

performing gray value closing operation of preset structural elements on the first coarse grid region, and then sequentially performing gray threshold segmentation, closing operation and opening operation to finally obtain the vertical grid line; the photovoltaic printing screen comprises grid lines and steel wires, wherein the grid lines are white, the steel wires are black, the purpose of gray value closing operation is to enable the white grid lines to completely cross over the black steel wires for connection, the purpose of gray threshold segmentation is to extract a coarse grid area, the purpose of closing operation is to adapt to the black grid blocking defect exceeding a preset area, the phenomenon that the extraction of vertical grid lines is influenced by the black grid blocking defect is avoided, and the purpose of opening operation is to extract the vertical grid lines.

In an optional example, the performing gray value difference on the vertical grid line by using the third coarse grid region to obtain a wave-shaped grid line preliminarily includes:

the vertical grid lines are expanded, then the expanded vertical grid lines are subjected to gray value difference by using the third coarse grid area, and the wave-shaped grid lines are obtained preliminarily.

In an optional example, the gray threshold segmentation adopted to obtain the third coarse grid region is a gray histogram threshold segmentation method or an extra large threshold segmentation method; the purpose of the gray threshold segmentation is to extract the part of the grid line with the highest gray value in the image.

In an optional example, the primarily obtained wavy grid line is shaped through skeleton extraction to obtain a final wavy grid line, specifically:

performing gray value threshold extraction in a first preset gray value range on the preliminarily obtained wavy grid lines to remove the influence of the background on the extraction of the wavy grid lines;

then removing the background influence, and performing skeleton extraction on the wavy grid line to obtain a wavy skeleton;

and expanding the wave-shaped framework to finish shaping the preliminarily extracted wave-shaped grid line to obtain the final wave-shaped grid line.

In an optional example, the extracting and combining of the gray scale features of the gate line region to obtain the gate line of the photovoltaic printing screen specifically includes:

extracting a region of the gray value in a second preset gray value range in the combined grid line region by utilizing gray value threshold segmentation to serve as an extraction region;

removing the extraction region from the combined grid line region through gray difference operation to obtain a residual region;

and filling the holes in the residual region to obtain the grid line of the photovoltaic printing screen.

In a second aspect, the present invention provides a photovoltaic screen printing plate defect detection system, including:

the image determining unit is used for determining images of the photovoltaic printing screen under the condition of two light sources; the two light sources are: the upper light source is positioned right above the photovoltaic printing screen and the lower light source is positioned right below the photovoltaic printing screen; the image under the upper light source is a reflection type illumination image, and the image under the lower light source is a transmission type illumination image; the photovoltaic printing screen plate may have defects, and the defects include: pinholes, blocked nets, dirt and broken filaments; the blocking net is arranged in the grid line area; breaking the steel wire of the broken wire finger printing screen;

the dynamic parameter measuring unit is used for measuring the dynamic parameters of the photovoltaic printing screen plate based on the lower light source image, and the dynamic parameters comprise: mesh width, steel wire width and steel wire inclination; the mesh width refers to the pixel width of a single grid in the photovoltaic printing screen;

the grid line extraction unit is used for performing Blob analysis on the lower light source image, extracting a continuous region with the highest gray value in the lower light source image, and taking the continuous region as a grid line of the photovoltaic printing screen;

a pinhole detection unit for removing the grid lines in the lower light source image based on the extracted grid lines and analyzing the highlight region information on the lower light source image from which the grid lines are removed to perform pinhole detection; when the area of a certain highlight area exceeds a preset area, the area is considered to be a pinhole correspondingly;

the device comprises a grid blocking detection unit, a grid blocking detection unit and a grid blocking detection unit, wherein the grid blocking detection unit is used for cutting a lower light source image by using extracted grid lines to generate a first gray scale image only containing the grid lines, extracting a blocking net in the first gray scale image only containing the grid lines by using Blob analysis, subtracting two gray scale images corresponding to an upper light source image and a lower light source image to obtain a subtracted gray scale image, cutting the subtracted gray scale image by using the extracted grid lines to generate a second gray scale image only containing the grid lines, performing Blob analysis on the second gray scale image only containing the grid lines, and further extracting blocking nets which are possibly missed to be detected in the two images; the gray scale image only containing the grid lines is a gray scale image of which the pixel values of other areas except the grid line area are 0;

the smudging detection unit is used for removing grid lines in the lower light source image and the upper light source image based on the extracted grid lines, and performing Blob analysis on the two images after the grid lines are removed based on the mesh width to detect the smudging of the screen printing plate;

and the broken wire detection unit is used for removing the grid lines in the lower light source image based on the extracted grid lines, detecting the steel wires in the lower light source image after the grid lines are removed according to the width of the steel wires and the gradient of the steel wires, and performing broken wire detection on the edges of the steel wires by utilizing Blob analysis.

In an optional example, the network blockage detection unit extracts an area of which the gray value is smaller than a first preset value; if the height of the extraction area is larger than the preset pixel and the area is larger than the mesh width, the area is considered as a blocking net at the joint of the grid line and the steel wire; if the area of the extraction region is larger than the mesh width and the radius of the inscribed circle is larger than the preset radius, the region is considered as a blocking net inside the grid line; and if the width of the extraction region is between the preset widths, the height is between the first preset heights, and the compactness is smaller than a threshold value, the region is considered as a grid blocking at the edge of the grid line.

In an optional example, after determining the gate lines in the image, the contamination detection unit sets, according to position information of the gate lines in the image, a region in the image, which is less than a preset distance from the gate lines, as a near gate region, and a region in the image, which is greater than the preset distance, as a far gate region; removing steel wires in the near-grid region according to the width of the steel wires and the gradient of the steel wires to obtain a first steel wire removing image, and segmenting the first steel wire removing image by using a gray threshold to extract a first suspected dirty region with a gray value smaller than a second preset value; if the height of the first suspected dirty area is between the second preset height and the area is larger than the mesh width, taking the first suspected dirty area as the dirty area of the near-grid area; removing steel wires in a far grid region according to the width of the steel wires and the gradient of the steel wires to obtain a second steel wire removing map, and segmenting a second steel wire removing image by using a gray threshold to extract a second suspected dirty region with a gray value smaller than a third preset value; the third preset value is a gray value corresponding to the trough after gray histogram statistics is carried out on the second type of steel wire removing image; and if the area of the second suspected dirty area is more than twice the mesh width, taking the second suspected dirty area as the dirty area of the far gate area.

In an optional example, the wire breakage detection unit performs wire breakage detection on the edge of the steel wire by using Blob analysis, specifically: if the edge of the steel wire is positioned in the image, the edge area is considered to be broken filaments; the image inner finger is a region with preset pixels away from the wide edge and the narrow edge of the image.

In an optional example, the grid line extracting unit extracts a continuous region with a highest gray value in a lower light source image, and uses the continuous region as a grid line of the photovoltaic printing screen, specifically: carrying out gray value closing operation on the lower light source image by using the rectangular structural elements, and then carrying out gray threshold segmentation to extract an area with a gray value larger than a preset value to obtain a first coarse grid area; performing gray value closing operation on the lower light source image by utilizing the hexagonal structural elements, and then performing gray threshold segmentation to extract an area with a gray value larger than a preset value to obtain a second coarse grid area; performing gray threshold segmentation on the lower light source image to obtain a third coarse grid region; wherein, the coarse grid region refers to a rough grid line region; sequentially carrying out gray value closing operation, gray threshold segmentation, closing operation and opening operation on the first coarse grid region to obtain a vertical grid line; carrying out gray value difference on the vertical grid lines by utilizing a third coarse grid region to obtain wave-shaped grid lines preliminarily, and shaping the wave-shaped grid lines preliminarily obtained through framework extraction to obtain final wave-shaped grid lines; carrying out gray value difference on the vertical grid line and the final wavy grid line by using the second coarse grid region to obtain a pattern grid line; the pattern grid lines refer to grid lines except vertical grid lines and wave-shaped grid lines; combining the vertical grid lines, the wave-shaped grid lines and the pattern grid lines to obtain combined grid line regions; and extracting the gray characteristic of the grid line region to obtain the grid line of the photovoltaic printing screen.

Generally, compared with the prior art, the above technical solution conceived by the present invention has the following beneficial effects:

the invention provides a method and a system for detecting defects of a photovoltaic printing screen plate, which can well solve the problems of low human eye detection efficiency and the need of reading Gerber dragging detection speed by autonomously extracting grid lines, assisting illumination by upper and lower light sources and designing an image processing algorithm aiming at different defect characteristics to detect the surface defects of the photovoltaic printing screen plate, and further improve the precision of the surface defect detection of the photovoltaic printing screen plate.

Drawings

Fig. 1 is a flowchart of a method for detecting defects of a photovoltaic printing screen according to an embodiment of the present invention;

FIG. 2 is a flowchart of a method for detecting surface defects of a Gerber-free photovoltaic screen printing plate according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of a system for detecting surface defects of a Gerber-free photovoltaic printing screen according to an embodiment of the present invention;

FIG. 4 is a partial enlarged view of a lower light source photograph of an exemplary photovoltaic printing screen provided in accordance with an embodiment of the present invention;

fig. 5 is a diagram illustrating a common screen blocking defect of a photovoltaic printing screen according to an embodiment of the present invention;

fig. 6 is a diagram illustrating a defect of a conventional stain on a photovoltaic screen printing plate according to an embodiment of the present invention;

fig. 7 is a diagram illustrating a defect of a conventional broken filament of a photovoltaic screen printing plate according to an embodiment of the present invention;

fig. 8 is a schematic diagram of a photovoltaic printing screen defect detection system according to an embodiment of the present invention;

the same reference numbers will be used throughout the drawings to refer to the same or like elements or structures, wherein: the device comprises an upper light source 1, a CCD camera 2, a lower light source 3, a three-dimensional moving platform 4, a screen clamping jig 5, a console 6 and a display screen 7.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the invention.

In order to solve the existing technical problems, the invention provides a Gerber-free photovoltaic screen printing plate detection method and system, so as to solve the problems that the human eye detection efficiency is low, the Gerber dragging detection speed needs to be read in the machine vision detection method, and the precision of the surface defect detection of the photovoltaic screen printing plate is further improved by adopting a method of an upper light source and a lower light source.

Fig. 1 is a flowchart of a method for detecting defects of a photovoltaic screen printing plate according to an embodiment of the present invention; as shown in fig. 1, the method comprises the following steps:

s101, determining images of the photovoltaic printing screen under two light sources; the two light sources are: the upper light source is positioned right above the photovoltaic printing screen and the lower light source is positioned right below the photovoltaic printing screen; the image under the upper light source is a reflection type illumination image, and the image under the lower light source is a transmission type illumination image; the photovoltaic printing screen plate may have defects, and the defects include: pinholes, blocked nets, dirt and broken filaments; the blocking net is arranged in the grid line area; the steel wire of the broken wire finger printing screen plate is broken;

s102, measuring dynamic parameters of the photovoltaic printing screen plate based on the lower light source image, wherein the dynamic parameters comprise: mesh width, steel wire width and steel wire inclination; the mesh width refers to the pixel width of a single grid in the photovoltaic printing screen;

s103, performing Blob analysis on the lower light source image, extracting a continuous region with the highest gray value in the lower light source image, and taking the continuous region as a grid line of the photovoltaic printing screen;

s104, removing grid lines in the lower light source image based on the extracted grid lines, and analyzing highlight area information on the lower light source image after the grid lines are removed to perform pinhole detection; when the area of a certain highlight area exceeds a preset area, the area is considered to be a pinhole correspondingly;

s105, cutting the lower light source image by using the extracted grid lines to generate a first gray scale image only containing the grid lines, analyzing and extracting the blocked nets in the first gray scale image only containing the grid lines by using Blob, subtracting two gray scale images corresponding to the upper light source image and the lower light source image to obtain a subtracted gray scale image, cutting the subtracted gray scale image by using the extracted grid lines to generate a second gray scale image only containing the grid lines, analyzing the Blob on of the second gray scale image only containing the grid lines, and further extracting the blocked nets which are possibly missed to be detected in the two images; the gray scale image only containing the grid lines is a gray scale image with the pixel value of 0 in other areas of the image except the grid line area;

s106, removing grid lines in the lower light source image and the upper light source image based on the extracted grid lines, and performing Blob analysis on the two images after the grid lines are removed based on the mesh width to detect the dirt of the screen printing plate;

and S107, removing the grid lines in the lower light source image based on the extracted grid lines, detecting the steel wires in the lower light source image after the grid lines are removed according to the width and the gradient of the steel wires, and performing broken wire detection on the edges of the steel wires by utilizing Blob analysis.

In some more specific embodiments, in one aspect, as shown in fig. 2, the present invention provides a method for online detecting surface defects of a Gerber-free photovoltaic screen printing plate, comprising the following steps:

and S1, collecting images. And shooting the regional image of the photovoltaic printing screen plate to be detected by using a CCD industrial camera clamped on the gantry platform, and sequentially shooting an upper light source picture and a lower light source picture in the same region each time.

And S2, preprocessing the image. And measuring partial dynamic parameters of the screen plate, such as mesh width, steel wire inclination and the like.

And S3, extracting the grid lines in the screen printing plate. The Blob analysis is performed using the lower light source picture taken in S1 to extract the grid lines.

S31: and carrying out gray value closing operation on the lower light source image.

Carrying out gray value closing operation on the lower light source image by using the rectangular structural elements, and then carrying out gray threshold segmentation to obtain a coarse grid region 1; carrying out gray value closed operation on the lower light source image by using the hexagonal structural elements, and then carrying out gray threshold segmentation to obtain a coarse grid region 2;

in this embodiment, the size of the rectangular structural elements is 15 × 15, and the subsequent grayscale threshold is 200-. The size of the hexagonal structuring element is 11 x 11, followed by a gray value threshold of 200-.

S32: and extracting the vertical grid line.

Carrying out gray value closing operation, gray threshold segmentation, closing operation and opening operation on the coarse grid region 1 to obtain a vertical grid line, specifically:

performing gray value closing operation of preset structural elements on the coarse grid region 1, and then sequentially performing gray threshold segmentation, closing operation and opening operation to finally obtain the vertical grid line; the photovoltaic screen printing plate comprises grid lines and steel wires, wherein the grid lines are white, the steel wires are black, the purpose of gray value closing operation is to enable the white grid lines to completely cross over the black steel wires for connection, the purpose of gray threshold segmentation is to extract a coarse grid area, the purpose of closing operation is to adapt to the black grid blocking defect exceeding a preset area, the phenomenon that the extraction of vertical grid lines is influenced by the black grid blocking defect is avoided, and the purpose of opening operation is to extract the vertical grid lines.

And performing square opening operation with the structural element of 1 × 50 on the coarse grid region 1 to obtain the accurate vertical grid line 1.

In order to adapt to the influence of the massive black grid blocking defect on the extraction of the white grid lines, before the opening operation is carried out, the closing operation with the structural element of 1 × 50 is carried out on the coarse grid region 1, and then the opening operation is carried out.

S33: and extracting the wavy grid lines.

Processing the collected original image by utilizing threshold segmentation to obtain a coarse grid area 3, and performing differential preliminary extraction on the wavy linear grid lines 1 after expanding the vertical grid lines 1 extracted in the step 3 by utilizing the coarse grid area 3.

In order to improve the accuracy of grid line extraction, the threshold segmentation here can use a gray histogram threshold segmentation method, a greater amount of threshold segmentation method, or the like. The main purpose is to extract the part of the grid line with the highest gray scale value in the screen plate.

And performing gray value threshold extraction with the parameter of 100-160 for the wave-shaped grid line 1 region, removing the influence of the background on the wave-shaped grid line extraction, then performing skeleton extraction on the wave-shaped grid line 1 to obtain a wave-shaped skeleton 1, and performing expansion on the basis of the extracted wave-shaped skeleton 1 to realize the shaping of the wave-shaped grid line to obtain a wave-shaped grid line 2, wherein the expansion parameter can be 3.5, so that the extraction precision of the grid line is improved.

S34: and extracting the pattern grid line.

And carrying out differential operation on the vertical grid lines 1 and the wavy grid lines 2 by using the coarse grid region 2 to accurately extract the pattern grid lines 1.

S35: combining the grid lines extracted in blocks, combining the vertical grid lines 1, the wave grid lines 2 and the pattern grid lines 1, and processing the combined grid lines by means of gray feature extraction and the like to obtain grid line areas of the screen printing plate.

The specific steps for processing the merged gate lines are as follows: 1. and extracting the region with the gray scale between 120 and 170 in the merging region by utilizing threshold segmentation. 2. These regions are removed from the merged region by a differential operation. 3. And finally, filling and shaping the area holes to obtain an accurate grid line area.

And S4, detecting and classifying the defects. And analyzing, extracting and positioning different surface defects by using the grid lines extracted in the S3 and the upper and lower light source pictures in the S1.

And S5, marking different defect information in the human-computer interaction interface by using different identifiers, so that the detection condition can be checked by the staff conveniently.

S6, after finishing the detection of the local surface defects, the CCD industrial camera and the upper light source are moved to the position of the next printing screen to repeat S2-S5 until the detection of one printing screen is finished.

The further step 1 specifically comprises:

and S11, installing the screen to be detected on the platform to be detected, initializing the shooting platform, and moving the camera to the initial position.

S12, placing a surface light source below the area to be detected, installing an annular light source on a clamp for clamping the CCD industrial camera, and shooting the upper light source picture and the lower light source picture alternately by the CCD industrial camera at the same position, namely only one light source is turned on when the CCD industrial camera works each time.

Furthermore, the upper light source shoots a reflective lighting picture when irradiating, and the protruded screen plate has depth information. The lower light source shoots a transmission type lighting picture during irradiation, and transparency transformation information in the screen printing plate is highlighted.

The CCD camera can be an area array black-and-white camera or a line array black-and-white camera.

The specific steps of step 4 are as follows:

and S41, detecting the pinhole at the edge of the grid line. And (3) in combination with the characteristics of good light transmission of the pinholes and consistent display effect and grid lines in the lower light source picture, removing the grid lines in the screen picture by using the grid lines in the step (2), and detecting the edge pinholes on the basis of the picture with the removed grid lines.

Judging whether the pinhole defect exists by judging whether the area of the highlight area of the screen picture with the grid lines removed exceeds a set threshold value, and positioning the pinhole defect by positioning the position of the highlight area.

The acquisition of the highlight area can adopt various threshold segmentation methods: histogram thresholding, gray-scale thresholding, Ostu thresholding, and the like.

And S42, detecting the net blocking defects. And (3) aiming at different grid blocking defects, positioning the grid lines in the upper and lower light source pictures in the step (1) by utilizing the grid lines in the step (3), reducing the light transmission of the grid lines by utilizing the grid blocking defects and changing the height information of the grid lines to detect the grid blocking defects.

A region with a gray value of less than 240 in the grid line region is extracted by performing Blob analysis on a lower light source picture, and whether the grid blocking defect is detected by using the characteristics of area, height, width, compactness and the like.

Meanwhile, the coverage of the algorithm to the defects is enlarged by subtracting the pictures of the upper and lower light sources, and certain grid blocking defects are more obvious than the gray characteristics in the independent upper and lower light sources after the subtraction of the upper and lower light sources. And designing a Blob analysis method for the subtracted pictures to realize detection of related network blockage defects.

Specifically, aiming at different net blocking defects, after Blob analysis is adopted: the feature threshold is set to: the height is more than 6 pixels, and the area is more than mesh width, so that the defect of net blocking which cannot be separated visually from the steel wire can be extracted; setting the characteristic threshold to: the area is larger than the mesh width, and the radius of the inscribed circle is larger than 1. The grid blocking defect of the grid line inside can be extracted; the feature threshold is set to: 2< width <10, 2< height 10, compactness <1.5 can extract the defect with smaller area at the edge of the grid line.

And S43, detecting the dirt defect. And (4) removing the grid lines in the upper and lower light source pictures by using the grid lines in the step (3), and analyzing and extracting dirty features by using Blob.

Specifically, the grid lines in the picture are removed by using the obtained grid line regions, and then the grid line region close to the grid lines and the grid line region far away from the grid lines are divided.

And (3) removing the steel wires in the near-grid region to obtain a steel wire removing image 1, segmenting the steel wire removing image 1 by using a gray threshold to extract a suspected dirty region 1 with the gray value smaller than 80, setting the characteristic threshold to be more than 3 and less than 20, and extracting the near-grid region dirty from the suspected dirty region 1 by using the area width.

And removing steel wires from the far-grid region to obtain a steel wire removing graph 2, and extracting a suspected dirty region between 0 and a threshold value 1 by using threshold segmentation, wherein the threshold value 1 is a gray value corresponding to a trough after gray histogram statistics is carried out on the steel wire removing graph 2. The threshold was set to area > mesh wide x 2 to extract far grid soiled areas from the suspected soiled area 2.

And S44, detecting the broken wire defect. And aiming at the edge of the broken steel wire, the edge of the picture can not exist, the picture with the grid lines removed is used for detecting the steel wire and then extracting the edge of the steel wire, whether the edge of the steel wire is at the edge of the picture is judged, if not, the broken steel wire is judged, and if so, the broken steel wire is not judged.

On the other hand, the invention also provides a surface defect detection system of the Gerber-free photovoltaic printing screen, which comprises the following steps: the device comprises a detection platform, an image acquisition unit, an image preprocessing unit, a grid line extraction unit, a defect detection unit and a defect display unit.

The detection platform and the image acquisition unit: as shown in fig. 3, an upper light source 1 and a lower light source 3 provide illumination, a CCD camera 2 captures images, a three-dimensional moving platform 4 can drive the camera and a clamp to move in three coordinate axes of XYZ, a screen clamping jig 5 fixes a clamping screen, a console 6 controls the three-dimensional moving platform and a display, and a display 7 displays the captured images of the screen and the defect detection results. The control platform 6 is connected with the moving platform 4 and the CCD camera 2, the CCD camera 2 and the upper light source 1 are clamped on a Z axis together, and the lower light source 3 is arranged below the screen clamping jig 5.

The light source can be a white light source, and different single-color LED light sources can also be adopted for screens with different colors.

The industrial camera may be a CCD area camera and a CCD line camera.

Image preprocessing unit: the device is used for processing the picture of the photovoltaic screen printing plate to be detected, which is acquired by the image acquisition unit, and measuring dynamic parameters including opening width, mesh width, steel wire inclination and the like.

Thirdly, a grid line extraction unit: the grid lines of the common photovoltaic printing screen are divided into vertical grid lines, wave-shaped grid lines and pattern grid lines, extraction is respectively carried out by means of Blob analysis, and finally the grid lines are combined together.

A defect detection unit: according to several common surface defects, performing Blob analysis, performing feature screening to extract different defects, and counting the number, types, areas and other features of the defects in the extraction process.

A defect display unit: the console displays the detection result and the analysis and statistics result on a display screen, so that the working personnel can conveniently process or repair the defects.

In order to facilitate understanding of the technical solutions of the present invention, the following specific examples are given to describe the technical solutions of the present invention in detail:

fig. 4 is a partial enlarged view of a typical lower light source photographed image of a screen printing plate, which is composed of a grid line, a steel wire and a latex layer.

The grid line is a white area in fig. 4, and is transparent, so the area is characterized by a high gray value in the figure and is visually bright white.

The steel wires are black in the grid of fig. 4, which is called steel wires, and are opaque, so that the gray value of the area in the graph is low, and the area is visually pure black.

The latex layer is a gray area in fig. 4, namely the latex layer, and the light transmittance is between the two areas.

The common defects of the photovoltaic printing screen are as follows:

1. blocking the net; this defect is most common and most important and directly affects the conductivity of the printed solar cell. The portion of the wire shown in fig. 5 is a screen defect. The non-steel wire parts of the white grid lines are shielded by the white grid lines and are called as net blocking.

2. Smudging; as shown by the frame line regions in the two small graphs (a) and (b) in fig. 6, the defect does not affect the gray value of the latex layer in the grid line region.

3. And the broken wire is the broken steel wire, as shown in the area of the frame line in fig. 7.

4. The pinholes, i.e., the latex layer is broken, light can penetrate through the pinholes, and the non-grid line areas also have places with particularly high gray values.

In a specific embodiment, measuring part of the dynamic parameters of the halftone specifically includes:

the measured parameters were:

average gray value of the acquired picture: and the method is used for judging whether the picture polishing condition meets the measurement requirement. And if the average gray value of the upper light source picture is between 60 and 160, the upper light source picture meets the measurement requirement, and if the average gray value of the lower light source picture is between 30 and 140, the lower light source picture meets the measurement requirement. Pictures that do not meet the measurement requirements need to be re-acquired.

The mesh width is as follows: defined as the pixel width of a single grid in the halftone.

The width of the steel wire.

Among them, mesh width and steel wire width are widely used as parameters in morphological operations for subsequent defect detection. For example, the mesh width is used as a parameter of morphological operation at multiple positions for detecting dirt and network blockage defects.

Specifically, the steel wire extraction method comprises the following specific steps: all black areas are screened according to the gray value of 0-100, and then a rectangular mask with the size (length and width) is used for carrying out open operation on the target area. Among them, the open operation is a common technical means in Blob analysis.

The extraction of horizontal steel wires can adopt a rectangular mask plate (the width of the steel wire is 2, 1) with the length larger than the width, and the extraction of vertical steel wires adopts a rectangular mask plate (the width of the steel wire is 2) with the width larger than the length.

More specifically, the steel wire inclination parameters are used in the steel wire extraction process. In the wire breakage detection, steel wires in a single direction are independently extracted for end point detection, for example, all steel wires in the horizontal direction are extracted at one time for end point detection, whether wire breakage exists in the horizontal direction is judged, and then the steel wires in the vertical direction are detected in the same step. Some types of screens are not 90 degrees across flat vertical, but are angled. This does not facilitate the extraction of steel wires in a single direction using the above method. For example, after the angle of the steel wire is measured, the picture can be rotated by the same angle, and the steel wire can be changed into a horizontal and vertical one. The steel wire can be extracted in one direction by the above method.

Fig. 8 is a schematic diagram of a photovoltaic screen printing plate defect detection system according to an embodiment of the present invention, as shown in fig. 8, including:

an image determining unit 810, configured to determine an image of the photovoltaic printing screen under two light sources; the two light sources are: the upper light source is positioned right above the photovoltaic printing screen and the lower light source is positioned right below the photovoltaic printing screen; the image under the upper light source is a reflection type illumination image, and the image under the lower light source is a transmission type illumination image; the photovoltaic printing screen plate may have defects, and the defects include: pinholes, blocked nets, dirt and broken filaments; the blocking net is arranged in the grid line area; the steel wire of the broken wire finger printing screen plate is broken;

a dynamic parameter measuring unit 820, configured to measure a dynamic parameter of the photovoltaic screen printing plate based on the lower light source image, where the dynamic parameter includes: mesh width, steel wire width and steel wire inclination; the mesh width refers to the pixel width of a single grid in the photovoltaic printing screen;

a grid line extraction unit 830, configured to perform Blob analysis on the lower light source image, extract a continuous region with a highest gray scale value in the lower light source image, and use the continuous region as a grid line of the photovoltaic printing screen;

a pinhole detecting unit 840 for removing the gate lines in the lower light source image based on the extracted gate lines, and analyzing the highlight region information on the lower light source image from which the gate lines are removed to perform pinhole detection; when the area of a certain highlight area exceeds a preset area, the area is considered to be a pinhole correspondingly;

a blocking detection unit 850, configured to cut the lower light source image by using the extracted gate lines, generate a first gray scale image only including the gate lines, extract a blocking net in the first gray scale image only including the gate lines by Blob analysis, subtract the two gray scale images corresponding to the upper light source image and the lower light source image to obtain a subtracted gray scale image, cut the subtracted gray scale image by using the extracted gate lines, generate a second gray scale image only including the gate lines, perform Blob analysis on the second gray scale image only including the gate lines, and further extract a blocking net which may be missed to be detected in the two images; the gray scale image only containing the grid lines is a gray scale image of which the pixel values of other areas except the grid line area are 0;

a dirt detection unit 860, configured to remove the grid lines in the lower light source image and the upper light source image based on the extracted grid lines, and perform Blob analysis on the two images after the grid lines are removed based on the mesh width to detect dirt on the halftone;

and a broken wire detection unit 870 for removing the grid lines in the lower light source image based on the extracted grid lines, detecting the steel wires in the lower light source image from which the grid lines are removed according to the steel wire width and the steel wire inclination, and performing broken wire detection on the steel wire edges by using Blob analysis.

It can be understood that detailed functional implementation of each unit in fig. 8 can refer to the description in the foregoing method embodiment, and is not described herein again.

It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.

Claims

1. A photovoltaic screen printing plate defect detection method is characterized by comprising the following steps:

removing grid lines in the lower light source image based on the extracted grid lines, and analyzing highlight area information on the lower light source image after the grid lines are removed to perform pinhole detection; when the area of a certain highlight area exceeds a preset area, the area is considered to be a pinhole;

2. The method according to claim 1, wherein extracting the occlusion in the image by Blob analysis is specifically:

if the width of the extraction region is between the preset widths, the height is between the first preset heights, and the compactness is smaller than the threshold value, the region is considered as the grid blocking edge of the grid line.

3. The method according to claim 1, characterized in that the Blob analysis of the image detects soiling of the screen, in particular:

after grid lines in an image are determined, according to position information of the grid lines in the image, setting a region in the image, which is less than a preset distance away from the grid lines, as a near grid region, and setting a region in the image, which is greater than the preset distance away from the grid lines, as a far grid region;

4. The method according to claim 1, characterized in that the steel wire edge is subjected to a wire breakage test by means of Blob analysis, in particular:

5. The method according to claim 1, wherein Blob analysis is performed on the lower light source image, and the continuous region with the highest gray scale value in the lower light source image is extracted and used as the grid line of the photovoltaic printing screen, specifically:

carrying out gray value closing operation on the lower light source image by using the rectangular structural elements, and then carrying out gray threshold segmentation to extract an area with a gray value larger than a preset value to obtain a first coarse grid area; performing gray value closing operation on the lower light source image by utilizing the hexagonal structural elements, and then performing gray threshold segmentation to extract an area with a gray value larger than a preset value to obtain a second coarse grid area; performing gray threshold segmentation on the lower light source image to obtain a third coarse grid region; wherein, the coarse grid region refers to a rough grid line region;

carrying out gray value difference on the vertical grid line and the final wavy grid line by using the second coarse grid region to obtain a pattern grid line; the pattern grid lines refer to grid lines except vertical grid lines and wave-shaped grid lines;

6. The utility model provides a photovoltaic printing half tone defect detecting system which characterized in that includes:

the image determining unit is used for determining images of the photovoltaic printing screen under the conditions of two light sources; the two light sources are: the upper light source is positioned right above the photovoltaic printing screen and the lower light source is positioned right below the photovoltaic printing screen; the image under the upper light source is a reflection type illumination image, and the image under the lower light source is a transmission type illumination image; the photovoltaic printing screen plate may have defects, and the defects include: pinholes, blocked nets, dirt and broken filaments; the blocking net is arranged in the grid line area; the steel wire of the broken wire finger printing screen plate is broken;

a pinhole detection unit for removing the grid lines in the lower light source image based on the extracted grid lines and analyzing highlight area information on the lower light source image from which the grid lines are removed to perform pinhole detection; when the area of a certain highlight area exceeds a preset area, the area is considered to be a pinhole correspondingly;

7. The system according to claim 6, wherein the mesh blocking detection unit extracts a region in the image having a gray value smaller than a first preset value; if the height of the extraction area is larger than the preset pixel and the area is larger than the mesh width, the area is considered as a blocking net at the joint of the grid line and the steel wire; if the area of the extraction region is larger than the mesh width and the radius of the inscribed circle is larger than the preset radius, the region is considered as a blocking net inside the grid line; and if the width of the extraction region is between the preset widths, the height is between the first preset heights, and the compactness is smaller than a threshold value, the region is considered as a grid blocking at the edge of the grid line.

8. The system according to claim 6, wherein after the contamination detection unit determines the grid lines in the image, according to the position information of the grid lines in the image, the region in the image which is less than a preset distance away from the grid lines is set as a near grid region, and the region which is more than the preset distance away from the grid lines is set as a far grid region; removing steel wires in the near-grid region according to the steel wire width and the steel wire inclination to obtain a first steel wire removing image, and segmenting the first steel wire removing image by using a gray threshold to extract a first suspected dirty region with the gray value smaller than a second preset value; if the height of the first suspected dirty area is between the second preset height and the area is larger than the mesh width, taking the first suspected dirty area as the dirty area of the near-grid area; removing steel wires in a far grid region according to the width of the steel wires and the gradient of the steel wires to obtain a second steel wire removing map, and segmenting a second steel wire removing image by using a gray threshold to extract a second suspected dirty region with a gray value smaller than a third preset value; the third preset value is a gray value corresponding to the trough after gray histogram statistics is carried out on the second type of steel wire removing image; and if the area of the second suspected dirty region is greater than twice the mesh width, then it is considered as dirty in the far gate region.

9. The system according to claim 6, wherein the wire breakage detection unit performs wire breakage detection on the edge of the steel wire by using Blob analysis, and specifically comprises: if the edge of the steel wire is positioned in the image, the edge area is considered to be broken filaments; the image inner finger is away from the area with preset pixels on the wide side and the narrow side of the image.

10. The system of claim 6, wherein the grid line extraction unit extracts a continuous region with the highest gray scale value in the lower light source image, and uses the continuous region as the grid line of the photovoltaic printing screen, specifically: carrying out gray value closing operation on the lower light source image by using the rectangular structural elements, and then carrying out gray threshold segmentation to extract an area with a gray value larger than a preset value to obtain a first coarse grid area; performing gray value closing operation on the lower light source image by utilizing the hexagonal structural elements, and then performing gray threshold segmentation to extract an area with a gray value larger than a preset value to obtain a second coarse grid area; performing gray threshold segmentation on the lower light source image to obtain a third coarse grid region; wherein, the coarse grid region refers to a rough grid line region; sequentially carrying out gray value closing operation, gray threshold segmentation, closing operation and opening operation on the first coarse grid region to obtain a vertical grid line; carrying out gray value difference on the vertical grid lines by utilizing a third coarse grid region to obtain wave-shaped grid lines preliminarily, and shaping the wave-shaped grid lines preliminarily obtained through framework extraction to obtain final wave-shaped grid lines; carrying out gray value difference on the vertical grid line and the final wavy grid line by using the second coarse grid region to obtain a pattern grid line; the pattern grid lines refer to grid lines except vertical grid lines and wave-shaped grid lines; combining the vertical grid lines, the wave-shaped grid lines and the pattern grid lines to obtain combined grid line regions; and extracting the gray characteristic of the grid line region to obtain the grid line of the photovoltaic printing screen.