CN112561875A - Photovoltaic cell panel coarse grid detection method based on artificial intelligence - Google Patents

Photovoltaic cell panel coarse grid detection method based on artificial intelligence Download PDF

Info

Publication number
CN112561875A
CN112561875A CN202011466964.7A CN202011466964A CN112561875A CN 112561875 A CN112561875 A CN 112561875A CN 202011466964 A CN202011466964 A CN 202011466964A CN 112561875 A CN112561875 A CN 112561875A
Authority
CN
China
Prior art keywords
measured
points
point
main grid
cell panel
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.)
Withdrawn
Application number
CN202011466964.7A
Other languages
Chinese (zh)
Inventor
孙猛猛
夏永霞
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Individual
Original Assignee
Individual
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Individual filed Critical Individual
Priority to CN202011466964.7A priority Critical patent/CN112561875A/en
Publication of CN112561875A publication Critical patent/CN112561875A/en
Withdrawn legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0004—Industrial image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/243—Classification techniques relating to the number of classes
    • G06F18/2433—Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
    • G06T5/70—
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/136—Segmentation; Edge detection involving thresholding
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/70—Determining position or orientation of objects or cameras
    • G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/48—Extraction of image or video features by mapping characteristic values of the pattern into a parameter space, e.g. Hough transformation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10004—Still image; Photographic image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10024—Color image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20048—Transform domain processing
    • G06T2207/20061—Hough transform
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20084—Artificial neural networks [ANN]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30108—Industrial image inspection

Abstract

The invention provides a photovoltaic cell panel coarse grid detection method based on artificial intelligence. Photovoltaic cell panel images are collected through the unmanned aerial vehicle, and cell panel images are extracted through the cell panel detection network. Performing threshold segmentation on the grayscale image of the cell panel and then performing opening operation to obtain a main grid image; converting the pixels with the gray value of 1 into Hough parameter space through coordinate transformation, and screening out points to be measured; matching the points to be measured according to the distance between the points to be measured and the m coordinate to obtain a set of the points to be measured, calculating the maximum distance between straight lines corresponding to the points to be measured in the set of the points to be measured aiming at any two sets of the points to be measured, and judging whether the set is a main grid point set or not according to the product of the maximum distances between the two sets of the points to be measured to obtain a main grid point set and outliers; and judging whether the outlier is a coarse grid point according to the m coordinate, and converting the coarse grid point into an image coordinate space to obtain the accurate position of the coarse grid.

Description

Photovoltaic cell panel coarse grid detection method based on artificial intelligence
Technical Field
The application relates to the field of computer vision, in particular to a photovoltaic cell panel coarse grid detection method based on artificial intelligence.
Background
During the printing plate making process of the photovoltaic cell panel, the situation of wrong printing may exist, specifically, broken grids, thick grids and the like. This kind of panel that has the wrong condition of printing, its generating efficiency can be than the generating efficiency low of normal printed cell board, and the printing problem can't change in the photovoltaic cell board use, consequently reports the condition of hindering because of the generating efficiency that the printing problem produced is different with the standard value, needs detect the coarse grid condition on the photovoltaic cell board.
At present, whether the photovoltaic cell panel has a coarse grid or not is detected usually by adopting a manual inspection mode, and the photovoltaic cell panel is observed by naked eyes of people and the photovoltaic cell panel with the coarse grid is recorded. The method has the problems that time and labor are consumed by manual inspection, and particularly, the workload is higher in the manual inspection in a large photovoltaic power station; and the condition of missed detection and false detection can exist due to strong subjectivity of judgment during manual inspection, and the condition of missed detection and false detection cannot be corrected at the moment if the coarse grid detection work is performed by manual inspection generally only once.
Disclosure of Invention
Aiming at the problems, the invention provides a photovoltaic cell panel coarse grid detection method based on artificial intelligence. Photovoltaic cell panel images are collected through the unmanned aerial vehicle, and cell panel images are extracted through the cell panel detection network. Performing threshold segmentation on the grayscale image of the cell panel and then performing opening operation to obtain a main grid image; converting the pixels with the gray value of 1 into Hough parameter space through coordinate transformation, and screening out points to be measured; matching the points to be measured according to the distance between the points to be measured and the m coordinate to obtain a set of the points to be measured, calculating the maximum distance between straight lines corresponding to the points to be measured in the set of the points to be measured aiming at any two sets of the points to be measured, and judging whether the set is a main grid point set or not according to the product of the maximum distances between the two sets of the points to be measured to obtain a main grid point set and outliers; and judging whether the outlier is a coarse grid point according to the m coordinate, and converting the coarse grid point into an image coordinate space to obtain the accurate position of the coarse grid.
A photovoltaic cell panel coarse grid detection method based on artificial intelligence is characterized by comprising the following steps:
step S1: acquiring an original image of a photovoltaic cell panel through a camera, detecting a photovoltaic cell panel area in the original image through a cell panel detection network, and outputting a cell panel image; carrying out binarization processing on the photovoltaic cell panel image, and then carrying out opening operation to obtain a binary image;
step S2: converting pixels with nonzero gray values in the binary image from an image coordinate space into a Hough parameter space, wherein the horizontal coordinate of the Hough parameter space is m, the vertical coordinate of the Hough parameter space is b, counting the number a of straight lines passing through each intersection point aiming at all straight line intersection points in the Hough parameter space, and screening out points to be measured;
step S3: matching the points to be measured according to the distance between the points to be measured to obtain a point set to be measured; calculating the maximum distance l of corresponding straight lines of the points to be measured in the set in the image coordinate spacemaxAiming at l corresponding to any two point sets to be measuredmaxMultiplying to obtain an area s, judging whether the point set to be measured is a main grid point set according to the proximity of the area s and the area s' of the image of the cell panel, and outputting the main grid point set;
step S4: and setting the central point of the main grid point as a main grid point, setting the rest points to be measured as outliers, judging whether the outliers are coarse grid points according to the consistency of the m coordinates of the outliers and the m coordinates of the main grid points, and transforming the coarse grid points into an image coordinate space to obtain the positions of the coarse grids.
The training method of the panel detection network comprises the following steps: taking a plurality of original images as a data set; manually marking the data set, marking out a surrounding frame of the photovoltaic cell panel area, and generating marking data; training is performed using a mean square error loss function.
The step of counting the number a of straight lines passing through each intersection point and screening out points to be measured comprises the following steps: setting a quantity threshold value beta, and judging the intersection point as a point to be measured when a is larger than or equal to beta; when a < beta, the intersection point is judged not to be the point to be measured.
The step of matching the points to be measured according to the distance between the points to be measured to obtain a set of the points to be measured comprises the following steps:
step S301: aiming at any two points p to be measured with the same m coordinate1、p2Forming a point set g to be measured, and connecting the line segments p1p2Is set to the minimum distance dmin；
Step S302: selecting a point p to be measured which does not belong to g3Calculating p1p3Length d of1,p2p3Length d of2；
If it is
Figure BDA0002834645940000021
And
Figure BDA0002834645940000022
are all positive integers and p3M coordinate and p1、p2When they are equal, p is3Adding a point set g to be measured; otherwise, p is not substituted3Adding a point set g to be measured;
step S303: and repeating the step S302 until all the points to be measured are traversed, and outputting a set g of the points to be measured.
The method for judging whether the point set to be measured is the main grid point set comprises the following steps:
converting the points to be measured in a point set to be measured into an image coordinate space to obtain a plurality of straight lines to be measured, calculating the distance l between every two straight lines to be measured, and selecting the largest l as the l corresponding to the point set to be measuredmax；
L ' corresponding to g ' is obtained through calculation aiming at any two to-be-measured point sets g ' and g ' with different m coordinates 'maxG' corresponding to lmaxFrom s ═ l'max×l″maxObtaining an area s, and calculating the area s' of the battery plate image;
setting an experience threshold gamma, and judging that the two point sets to be measured are not main grid point sets when s' -s is more than or equal to gamma; when s' -s is less than gamma, judging that the two point sets to be measured are a pair of main grid point sets to be measured;
and counting the total number d of main grid points contained in each pair of main grid point sets to be determined, setting a pair of main grid point sets with the maximum d as a main grid point set, and setting the points to be measured in the main grid point set as the main grid points.
The judging whether the outlier is a coarse grid point according to whether the m coordinate of the outlier is consistent with the m coordinate of the main grid point comprises the following steps: comparing the m coordinate value of the outlier with the m coordinate value of the main grid point in the main grid point set, and if the m coordinate values are consistent, judging that the outlier is a coarse grid point; if the m coordinate values are not consistent, the outlier is judged to be a non-coarse grid point.
Compared with the prior art, the invention has the following beneficial effects:
(1) the pixels belonging to the photovoltaic cell panel are detected through the neural network, the image of the photovoltaic cell panel can be accurately segmented, and the method is high in precision and speed.
(2) The influence of noise and normal fine grids is removed through the opening operation, and the position of the grid line can be detected more accurately.
(3) The main grid points are detected by utilizing the parallel property of the grid lines, so that the influence of scratches on the grid line detection can be avoided, and the undetermined main grid point set is screened according to the area of the battery panel image, so that the false detection is avoided.
(4) And screening coarse grid points from the outliers according to whether the m coordinate is consistent with the main grid point, so that the position of the coarse grid can be more accurately positioned.
Drawings
FIG. 1 is a process flow diagram.
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 are not intended to limit the invention.
It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
The first embodiment is as follows:
the invention mainly aims to realize the detection of the coarse grid defect of the photovoltaic cell panel.
In order to realize the content of the invention, the invention designs a photovoltaic cell panel coarse grid detection method based on artificial intelligence, and a flow chart of the method is shown in figure 1.
The process of printing and plate making of the photovoltaic cell panel comprises a screen printing process. After the screen printing process, converging and diverging metal printing contacts, namely a main grid and a fine grid, are formed on the surface of the photovoltaic cell. The main grid is connected directly to the external leads, while the finer metallization areas collect the current for delivery to the cell bus. Defects in the main gate and the fine gate cause an increase in resistance and a decrease in current transfer efficiency. Some screen printing defects are difficult to observe and detect by naked eyes, and special calipers are required for measurement. The printed defects comprise thick grids, and the width of the grid lines on the photovoltaic cell panel is larger than that of the standard grid lines. The invention discloses a method for detecting whether a main grid has a coarse grid defect.
Step S1:
set up the RGB camera on unmanned aerial vehicle, unmanned aerial vehicle patrols and examines the in-process and utilize the RGB camera to shoot the original image of photovoltaic cell board, original image RGB image. Should set up reasonable height for unmanned aerial vehicle. In order to enable the obtained original image to at least contain a whole photovoltaic cell panel single plate, the height of the unmanned aerial vehicle is not low. The original image comprises the photovoltaic cell panel and the area around the photovoltaic cell panel, and the method only needs the image of the area of the photovoltaic cell panel, so that the image of the area of the photovoltaic cell panel is segmented by using a cell panel detection network.
The training method of the panel detection network comprises the following steps: taking a plurality of original images as a data set; manually labeling the data set, labeling a surrounding frame of a photovoltaic cell panel area in an original image, and generating labeled data; the training of the network is performed using a mean square error loss function.
Inputting the original image into a trained battery panel detection network, detecting a photovoltaic battery panel area, outputting a surrounding frame of the photovoltaic battery panel area, cutting the original image, reserving the photovoltaic battery panel area, removing objects around the photovoltaic battery panel, and obtaining a battery panel image.
The invention detects whether the main grid has the defect of a coarse grid, and processes the battery plate image as follows in order to eliminate the influence of the fine grid on the detection of the coarse grid: and converting the battery panel image into a gray image, and performing binarization processing on the gray image by using a self-adaptive threshold segmentation algorithm to obtain a binary image.
The colors of the photovoltaic cell panels in the cell panel images are obviously distinguished, the grid lines are white, the cell panels are blue, however, due to the influence of factors such as illumination and the like, binarization processing cannot be performed by adopting a fixed threshold value, and therefore, the gray level images are segmented by adopting an adaptive threshold value segmentation algorithm. Adaptive threshold segmentation algorithms are well known and are not intended to be within the scope of the present invention, and the implementer may select an appropriate algorithm based on the circumstances. Such as the greater jin threshold method, the iterative method, etc. And (4) obtaining a binary image by segmentation, wherein the pixel value of the grid line pixel in the binary image is 1, and the pixel values of the rest pixels are 0.
In order to eliminate the influence of the fine grid, the binary image needs to be morphologically processed, and the method adopts open operation, namely, the binary image is subjected to corrosion operation and then expansion operation, so that discrete bright spots and the fine grid in the binary image can be eliminated. And obtaining a main grid image, wherein the main grid image comprises a main grid and straight lines generated by scratches.
Step S2:
carrying out straight line detection on the main grid image by using Hough transform, and specifically comprising the following steps of: traversing pixel points with pixel values of 1 in the main grid image, and counting the coordinates (x) of each point with pixel value of 1i,yi) Each pixel value is addedAnd converting the point which is 1 into a Hough parameter space from an image coordinate space, wherein the horizontal coordinate of the Hough parameter space is m, and the vertical coordinate of the Hough parameter space is b. The mathematical model of coordinate space transformation is b ═ mxi+yiNamely, one point in the image coordinate space corresponds to one straight line in the Hough parameter space. Similarly, a point in the hough parameter space corresponds to a straight line in the image coordinate space. It should be noted that, when a straight line in the image coordinate space is perpendicular to the x-axis, the points on the straight line are converted to corresponding straight lines in the hough parameter space and are parallel to each other, so that the image coordinate space is established such that the main grid is not perpendicular to the x-axis and is not parallel to the x-axis.
Obtaining a plurality of straight lines in the Hough parameter space through coordinate transformation, wherein the straight lines have a plurality of intersection points, detecting the number a of the straight lines passing through each straight line intersection point, setting an empirical number threshold beta, and when a is more than or equal to beta, indicating that the number of points of the intersection point corresponding to the straight line on which the pixel is 1 in the image coordinate space is enough, and judging the intersection point as a point to be measured; when a < beta, the intersection point is judged not to be the point to be measured. And outputting a plurality of points to be measured, wherein the points to be measured correspond to a straight line in the image coordinate space.
And recording the point to be measured as p, wherein the point to be measured which is the main grid corresponding to the straight line in the image coordinate space needs to be screened out, and the point to be measured is called as the main grid point. The main grids on the photovoltaic cell panel detected by the invention are crossed, the corresponding straight lines have two slopes, the straight lines with different slopes are mutually vertical, the straight lines with the same slopes are parallel, and the distances between the adjacent straight lines are equal. Considering that the adjacent main grids are in certain distance and are parallel to each other, the invention sets the following rules to match the points to be measured:
step S301: aiming at any two points p to be measured with the same m coordinate1、p2Forming a point set g to be measured, and connecting the line segments p1p2Is set to the minimum distance dmin。
Step S302: selecting a point p to be measured which does not belong to g3Calculating p1p3Length d of1,p2p3Length d of2。
If d is1/dminAnd d2/dminAre all positive integers and p3M coordinate and p1、p2When they are equal, p is3Adding a point set g to be measured; otherwise, p is not substituted3And adding a point set g to be measured.
Step S303: and repeating the step S302 until all the points to be measured are traversed, and outputting a set g of the points to be measured.
The main grid has two slopes, so that the straight line corresponding to the point to be measured in the image coordinate space in the two point to be measured sets is the main grid, but due to scratch interference and p1、p2And if the selection is improper, more than two to-be-measured point sets g are obtained.
In order to determine whether a straight line corresponding to a midpoint of a point set g to be measured in an image coordinate space is a main grid, the invention sets the following rules for judgment:
converting the points to be measured in a point set to be measured into an image coordinate space to obtain a plurality of straight lines to be measured, calculating the distance l between every two straight lines to be measured, and selecting the largest l as the l corresponding to the point set to be measuredmax；
L ' corresponding to g ' is obtained through calculation aiming at any two to-be-measured point sets g ' and g ' with different m coordinates 'maxG' corresponding to lmaxFrom s ═ l'max×l″maxObtaining an area s, and calculating the area s' of the battery plate image;
setting an experience threshold gamma, and judging that the two point sets to be measured are not main grid point sets when s' -s is more than or equal to gamma; and when s' -s is less than gamma, judging that the two to-be-detected point sets are a pair of to-be-determined main grid point sets.
And counting the total number d of the points to be measured contained in each pair of main grid point sets to be determined, and setting a pair of main grid point sets with the maximum d as a main grid point set. And setting the point to be measured in the main grid point set as a main grid point.
And setting the points to be measured except the main grid points as outliers, wherein the outliers comprise non-rough grid points corresponding to the scratch and rough grid points corresponding to the part of the rough grid which is more than the normal main grid. Screening coarse grid points in the outliers according to the m coordinates, setting the m coordinates of the main grid points in the two main grid point sets as m 'and m', and if the m coordinate of the outlier is equal to m 'or m', judging that the points are the coarse grid points and are coarse grids caused by printing errors; if the m coordinate of the outlier is not equal to m 'or m', the point is determined to be a non-coarse grid point and is a straight line generated by scratches or foreign matters.
According to the coordinates of the coarse grid points in the Hough parameter space, a linear equation of the coarse grid points corresponding to the coarse grid in the image coordinate space can be obtained, the positions with the coarse grid defects can be accurately positioned, and the subsequent processing of the coarse grid is facilitated.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.

Claims (6)

1. A photovoltaic cell panel coarse grid detection method based on artificial intelligence is characterized by comprising the following steps:
step S1: acquiring an original image of a photovoltaic cell panel through a camera, detecting a photovoltaic cell panel area in the original image through a cell panel detection network, and outputting a cell panel image; carrying out binarization processing on the photovoltaic cell panel image, and then carrying out opening operation to obtain a binary image;
step S2: converting pixels with nonzero gray values in the binary image from an image coordinate space into a Hough parameter space, wherein the horizontal coordinate of the Hough parameter space is m, the vertical coordinate of the Hough parameter space is b, counting the number a of straight lines passing through each intersection point aiming at all straight line intersection points in the Hough parameter space, and screening out points to be measured;
step S3: matching the points to be measured according to the distance between the points to be measured to obtain a point set to be measured; calculating the maximum distance l of corresponding straight lines of the points to be measured in the set in the image coordinate spacemaxAiming at l corresponding to any two point sets to be measuredmaxMultiplying to obtain an area s, judging whether the point set to be measured is a main grid point set according to the proximity of the area s and the area s' of the image of the cell panel, and outputting the main grid point set;
step S4: and setting the central point of the main grid point as a main grid point, setting the rest points to be measured as outliers, judging whether the outliers are coarse grid points according to the consistency of the m coordinates of the outliers and the m coordinates of the main grid points, and transforming the coarse grid points into an image coordinate space to obtain the positions of the coarse grids.
2. The method of claim 1, wherein the training method of the panel network comprises:
taking a plurality of original images as a data set;
manually marking the data set, marking out a surrounding frame of the photovoltaic cell panel area, and generating marking data;
training is performed using a mean square error loss function.
3. The method of claim 1, wherein the counting the number a of straight lines passing through each intersection point and screening out points to be measured comprises:
setting a quantity threshold value beta, and judging the intersection point as a point to be measured when a is larger than or equal to beta; when a < beta, the intersection point is judged not to be the point to be measured.
4. The method of claim 1, wherein the matching of the points to be measured according to the distances between the points to be measured to obtain the set of points to be measured comprises:
step S301: aiming at any two points p to be measured with the same m coordinate1、p2Forming a point set g to be measured, and connecting the line segments p1p2Is set to the minimum distance dmin；
Step S302: selecting a point p to be measured which does not belong to g3Calculating p1p3Length d of1，p2p3Length d of2；
If it is
Figure FDA0002834645930000011
And
Figure FDA0002834645930000012
are all positive integers and p3M coordinate and p1、p2When they are equal, p is3Adding a point set g to be measured; otherwise, it will notp3Adding a point set g to be measured;
step S303: and repeating the step S302 until all the points to be measured are traversed, and outputting a set g of the points to be measured.
5. The method as claimed in claim 1, wherein the method for judging whether the point set to be measured is a main grid point set comprises the following steps:
converting the points to be measured in a point set to be measured into an image coordinate space to obtain a plurality of straight lines to be measured, calculating the distance l between every two straight lines to be measured, and selecting the largest l as the l corresponding to the point set to be measuredmax；
L ' corresponding to g ' is obtained through calculation aiming at any two to-be-measured point sets g ' and g ' with different m coordinates 'maxG' corresponding to lmaxFrom s ═ l'max×l″maxObtaining an area s, and calculating the area s' of the battery plate image;
setting an experience threshold, and judging that the two point sets to be measured are not main grid point sets when s' -s is more than or equal to gamma; when s' -s is less than gamma, judging that the two point sets to be measured are a pair of main grid point sets to be measured;
and counting the total number d of main grid points contained in each pair of main grid point sets to be determined, setting a pair of main grid point sets with the maximum d as a main grid point set, and setting the points to be measured in the main grid point set as the main grid points.
6. The method of claim 1, wherein determining whether an outlier is a coarse grid point based on whether the m-coordinate of the outlier is consistent with the m-coordinate of the main grid point comprises:
comparing the m coordinate value of the outlier with the m coordinate value of the main grid point in the main grid point set, and if the m coordinate values are consistent, judging that the outlier is a coarse grid point; if the m coordinate values are not consistent, the outlier is judged to be a non-coarse grid point.
CN202011466964.7A 2020-12-14 2020-12-14 Photovoltaic cell panel coarse grid detection method based on artificial intelligence Withdrawn CN112561875A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011466964.7A CN112561875A (en) 2020-12-14 2020-12-14 Photovoltaic cell panel coarse grid detection method based on artificial intelligence

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011466964.7A CN112561875A (en) 2020-12-14 2020-12-14 Photovoltaic cell panel coarse grid detection method based on artificial intelligence

Publications (1)

Publication Number Publication Date
CN112561875A true CN112561875A (en) 2021-03-26

Family

ID=75064404

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011466964.7A Withdrawn CN112561875A (en) 2020-12-14 2020-12-14 Photovoltaic cell panel coarse grid detection method based on artificial intelligence

Country Status (1)

Country Link
CN (1) CN112561875A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113674150A (en) * 2021-07-27 2021-11-19 上海洪朴信息科技有限公司 Photovoltaic appearance assembly distance measuring method
CN114355977A (en) * 2022-01-04 2022-04-15 浙江大学 Tower type photo-thermal power station mirror field inspection method and device based on multi-rotor unmanned aerial vehicle

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113674150A (en) * 2021-07-27 2021-11-19 上海洪朴信息科技有限公司 Photovoltaic appearance assembly distance measuring method
CN114355977A (en) * 2022-01-04 2022-04-15 浙江大学 Tower type photo-thermal power station mirror field inspection method and device based on multi-rotor unmanned aerial vehicle
CN114355977B (en) * 2022-01-04 2023-09-22 浙江大学 Tower type photo-thermal power station mirror field inspection method and device based on multi-rotor unmanned aerial vehicle

Similar Documents

Publication Publication Date Title
CN109870461B (en) Electronic components quality detection system
CN110400290B (en) Method and device for detecting chromatic aberration of solar cell and storage medium
TW201702586A (en) Optical film defect detection method and system thereof
CN111667475B (en) Machine vision-based Chinese date grading detection method
CN113506286A (en) Microwave chip defect detection method based on small sample data set of YOLOv5 algorithm
CN110648330B (en) Defect detection method for camera glass
CN112561875A (en) Photovoltaic cell panel coarse grid detection method based on artificial intelligence
CN116630304B (en) Lithium battery mold processing detection method and system based on artificial intelligence
CN116703911B (en) LED lamp production quality detecting system
CN113034488A (en) Visual detection method of ink-jet printed matter
CN115272350A (en) Method for detecting production quality of computer PCB mainboard
CN115272336A (en) Metal part defect accurate detection method based on gradient vector
CN112686890A (en) Artificial board surface defect detection method based on singular value decomposition
CN117152161B (en) Shaving board quality detection method and system based on image recognition
CN114155226A (en) Micro defect edge calculation method
CN112381794A (en) Printing defect detection method based on deep convolution generation network
CN108898584B (en) Image analysis-based full-automatic veneered capacitor welding polarity discrimination method
CN108898080B (en) Ridge line neighborhood evaluation model-based crack connection method
CN115170476B (en) Printed circuit board defect detection method based on image processing
CN116109577A (en) Printing label defect detection system and method
CN111832565B (en) Decision tree-based nixie tube identification method
CN112801963B (en) Video image occlusion detection method and system
CN113763359A (en) Intelligent detection and sensing system for metal near-surface defect detection
CN107123105A (en) Images match defect inspection method based on FAST algorithms
CN117455917B (en) Establishment of false alarm library of etched lead frame and false alarm on-line judging and screening method

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
WW01 Invention patent application withdrawn after publication

Application publication date: 20210326

WW01 Invention patent application withdrawn after publication