CN114022415A - Liquid crystal display defect detection method based on single-pixel feature clustering cluster establishment - Google Patents
Liquid crystal display defect detection method based on single-pixel feature clustering cluster establishment Download PDFInfo
- Publication number
- CN114022415A CN114022415A CN202111204952.1A CN202111204952A CN114022415A CN 114022415 A CN114022415 A CN 114022415A CN 202111204952 A CN202111204952 A CN 202111204952A CN 114022415 A CN114022415 A CN 114022415A
- Authority
- CN
- China
- Prior art keywords
- cluster
- pixel
- pixel points
- feature
- pixel point
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000007547 defect Effects 0.000 title claims abstract description 96
- 239000004973 liquid crystal related substance Substances 0.000 title claims abstract description 37
- 238000001514 detection method Methods 0.000 title claims abstract description 28
- 239000013598 vector Substances 0.000 claims abstract description 37
- 238000000605 extraction Methods 0.000 claims abstract description 35
- 238000000034 method Methods 0.000 claims abstract description 23
- 238000004364 calculation method Methods 0.000 claims abstract description 9
- 239000011159 matrix material Substances 0.000 claims description 38
- 238000001914 filtration Methods 0.000 claims description 9
- 230000002950 deficient Effects 0.000 claims description 7
- 238000005070 sampling Methods 0.000 claims description 6
- 239000013256 coordination polymer Substances 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000013135 deep learning Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 201000004569 Blindness Diseases 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 230000004393 visual impairment Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
- G06T7/0006—Industrial image inspection using a design-rule based approach
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30121—CRT, LCD or plasma display
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Testing Of Optical Devices Or Fibers (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
- Liquid Crystal (AREA)
Abstract
The invention provides a liquid crystal display defect detection method for establishing a cluster based on single-pixel characteristic clustering, which comprises the steps of firstly, obtaining an RGB three-channel image of a liquid crystal display, and also comprises the step of S1, obtaining gray values and r, g and b channel values of each pixel point of the image, and converting the image into a gray image; s2, performing traversal calculation on each pixel point in the gray-scale image, and generating a feature vector of each pixel point through a multi-mode and multi-scale feature extraction kernel; s3, judging whether the liquid crystal screen has defects or not according to the feature vector clustering result and the coordinate distribution condition of the pixel points in the clustered cluster; by adopting the method, the defect samples do not need to be collected, the method is more practical in the algorithm landing application, and the efficiency of the defect detection algorithm is effectively improved due to the small scale of the parameters; in addition, compared with the traditional algorithms such as local pixel values and the like, the defect detection provided by the method is more comprehensive.
Description
Technical Field
The invention relates to the technical field of liquid crystal display detection, in particular to a liquid crystal display defect detection method based on single-pixel feature clustering cluster establishment.
Background
Mobile technology devices, such as mobile phones, VR (Virtual Reality) devices, and the like, have not yet developed display screens. In production, the defect detection of the display screen is an essential step. The defects of the liquid crystal display screen are of various types and are very difficult to detect.
In contrast, the conventional detection method is manual detection, which has high requirements on inspectors, and the defects of the inspectors are detected manually for a long time, so that the eyes of the inspectors are greatly damaged, such as vision loss; in addition, the inspector works for a long time and is tired, so that the inspection omission is caused, and the probability that defective products flow into the market is increased.
In order to overcome the defects of manual detection, a defect analysis method based on image analysis is also provided; for example, in the method, the system and the device for detecting the defect of the liquid crystal screen based on the small sample deep learning of the chinese patent with the publication number CN109242829A, the defect detection model is trained by collecting the defect sample, but the occurrence of the defect is unknown, and then the color feature, the pixel size and the shape after the defect imaging are unknown, so that even if the defect detection model can be obtained based on the small sample training, the generalization capability of the model is limited; also, for example, chinese patent publication No. CN104978748A discloses a method for detecting defects of a liquid crystal panel based on local pixel values, which detects defects by calculating local pixels, but the local pixel calculation is based on the premise that an image is divided into mesh-shaped pixel blocks to perform local observation, but if the defect area is large, abnormality cannot be observed in the local mesh-shaped pixel blocks.
In view of the defects in the prior art, the difficulty of visual detection of the liquid crystal display is analyzed, and the defects mainly comprise the defects with no rule and weak difference; the defect forms (such as defect color, defect outline and defect zigzag) are irregular; the defect types of the same liquid crystal screen can be various; aiming at the detection difficulty, the invention provides a novel liquid crystal display defect detection method, so that the defect detection with better accuracy can be realized while the problems are solved.
Disclosure of Invention
The invention aims to provide a liquid crystal display defect detection method for establishing a cluster based on single-pixel characteristic clustering, which does not need to collect defect samples, is more practical in algorithm floor application, and effectively improves the efficiency of a defect detection algorithm due to small scale of parameters; in addition, compared with the traditional algorithms such as local pixel values and the like, the defect detection provided by the method is more comprehensive.
The embodiment of the invention is realized by the following technical scheme:
in a first aspect, a method for detecting defects of a liquid crystal display based on single-pixel feature clustering cluster establishment is provided, which includes the following steps:
s1, obtaining gray values and r, g and b channel values of all pixel points of the image, and converting the image into a gray map;
s2, performing traversal calculation on each pixel point in the gray-scale image, and generating a feature vector of each pixel point through a multi-mode and multi-scale feature extraction kernel;
s3, dividing the pixel points into two first clusters according to the result of the feature vector clustering, filtering the clusters to generate a second cluster, and if the second cluster is generated and the number of the pixel points in the second cluster is greater than a first threshold value, possibly having defects; if a second cluster is not generated and the difference value of the pixel points between the two first clusters is greater than a second threshold value, a defect may exist; and under the condition that defects possibly exist, judging whether the coordinate distribution of the pixel point of one first cluster or the second cluster with smaller pixel points in the first cluster is discrete, if so, judging that the first cluster or the second cluster is not defective, otherwise, judging that the first cluster or the second cluster is defective.
Further, the traversing calculation of each pixel point in the gray-scale map specifically includes:
by pixel points
Taking the local image area matrix with the same size as the characteristic extraction kernel as the center through sampling
Matrix multiplication or matrix subtraction is carried out;
the feature extraction kernel is a two-dimensional matrix f with the size of m × mm*m；
Wherein, the matrix multiplication operation is the multiplication of the two-dimensional matrix of the feature extraction kernel and the local image area matrix, as the following formula (1),
the matrix subtraction is to subtract a local image area matrix from a feature extraction kernel two-dimensional matrix, as shown in the following formula (2),
wherein,
to the pixel point
The local image area matrix obtained for center sampling
And the feature extraction kernel two-dimensional matrix fm*mThe result after multiplication or subtraction.
Further, the generating of the feature vector of the pixel point by the multi-modal and multi-scale feature extraction kernel specifically includes:
the multi-scale value, namely the size m of the feature extraction kernel, comprises a plurality of values, and the multi-scale value is the feature value in the feature extraction kernel
The definition of (a) includes a plurality;
and generating the feature vector of the pixel point according to the multi-modal feature value definition and different values of the multi-scale feature extraction kernel
Further, the S3 specifically includes:
s31, selecting one pixel point as a clustering initial center, traversing all other pixel points of the image, and calculating the Euclidean distance between the clustering initial center and the feature vectors of all other pixel points;
s32, screening out two pixel points with the maximum Euclidean distance from the characteristic vector of the clustering initial center as second clustering centers a and b;
s33, respectively calculating characteristic vector Euclidean distances between each pixel point and the second clustering center a and the second clustering center B, and classifying all the pixel points to obtain a first cluster A of the second clustering center a and a first cluster B of the second clustering center B;
s34, filtering the first cluster A and the first cluster B to generate a second cluster C, and judging whether defects exist or not according to whether the second cluster C is generated or not and the number of pixel points contained in the second cluster C;
and S35, according to the judgment result of the S34, judging whether the defect exists or not by combining the coordinate distribution of the pixel points in the first cluster A, the first cluster B and the second cluster C.
Further, the classifying all the pixel points in S33 specifically includes:
and when the Euclidean distance between the pixel point and the characteristic vector of the second clustering center a is greater than that between the pixel point and the characteristic vector of the second clustering center B, classifying the pixel point as a first cluster A, otherwise, classifying the pixel point as a first cluster B.
Further, the filtering process in S34 specifically includes:
calculating characteristic vector Euclidean distances between all pixel points in the first cluster A and the second cluster center a, and calculating and obtaining a characteristic vector Euclidean distance mean value dAThe characteristic vector Euclidean distance between the first cluster center a and the second cluster center a is larger than j x dAThe pixel points of (a) are deleted from the first cluster a and added to the second cluster C, where j is a constant term.
Further, the specific step of judging whether a defect may exist according to whether the second cluster C is generated or not and the number of pixels included in the second cluster C is as follows:
when a second cluster C is generated and the number of pixel points in the second cluster C is more than 5, judging that a defect possibly exists;
when the second cluster C is not generated, whether the difference of the pixel point numbers of the first cluster A and the first cluster B is larger than 1/3 of the total pixel point number of the image is judged, and if the difference is larger than 1/3, the defect is judged to exist.
Further, the S35 specifically includes:
under the condition that a second cluster C appears and the number of pixel points in the second cluster C is more than 5, calculating whether the coordinate distribution of the pixel points in the second cluster C is discrete or not; under the condition that the second cluster C does not appear, selecting a cluster with a small number of pixel points in the first cluster A and the first cluster B, defining the cluster as a cluster Q, and calculating whether the pixel coordinate distribution of the pixel points in the cluster Q is discrete or not; and if the pixel point coordinates are distributed in a discrete manner, judging that no defect exists, and if the pixel point coordinates are distributed in an independently polymerizable region, judging that the defect occurs.
Further, whether the coordinate distribution of the pixel points is discrete specifically is:
setting a pixel point coordinate distribution distance threshold d, wherein the distance threshold
Wherein IwRepresenting the pixel width, I, of an imagehIndicating that the image pixel is high; sequentially traversing the pixel points in the cluster Q or the second cluster C, and calculating the coordinate distribution distance value d of every two pixel points in the cluster Q or the second cluster CPIf d isPD is less than or equal to d, the two pixel points are the coordinate cluster set of the similar pixel points, and the total number N of the newly clustered coordinate distribution clusters is obtainedd；
Presetting a third threshold value T, wherein T is k NtWhere k is a constant term, NtThe total number of pixel points of the cluster Q or the second cluster C is NdWhen the number of the pixels in the cluster Q is larger than or equal to T, the pixel point coordinate distribution of the second cluster C is considered to be discrete, and the pixel point coordinate distribution is judged to be free of defects; otherwise, the coordinate distribution is considered to be in the independently polymerizable region, and the defect is judged.
In a second aspect, an electronic device is provided that includes a memory and a processor;
the memory stores computer-executable instructions;
the processor executes the computer-executable instructions stored in the memory, so that the processor executes the liquid crystal display defect detection method for establishing the cluster based on the single-pixel characteristic clustering.
The technical scheme of the embodiment of the invention at least has the following advantages and beneficial effects:
(1) compared with the existing deep learning technical scheme based on small samples, the method does not need to collect the defect samples, is more practical in algorithm floor application, and effectively improves the efficiency of the defect detection algorithm due to small scale of parameters;
(2) compared with the traditional algorithms such as local pixel values and the like, the defect detection is more comprehensive, and can be used for detecting the zigzag defects and large-area defects of the liquid crystal display.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. The components of embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.
In a first aspect, a method for detecting defects of a liquid crystal display based on single-pixel feature clustering cluster establishment is provided, and first, an RGB three-channel image of the liquid crystal display is obtained, which further includes the following steps:
s1, obtaining the gray value and the r, g and b channel values of each pixel point of the image, and converting the image into a gray map.
The general formula for converting an RGB three-channel image into a gray scale map is as follows,
wherein,
representing the gray value of the nth pixel point of the image,
respectively representing the r, g and b channel values of the nth pixel point of the image.
Generally, an RGB three-channel image of a liquid crystal panel is an image when the liquid crystal panel displays a single color.
And S2, performing traversal calculation on each pixel point in the gray-scale image, and generating a feature vector of each pixel point through multi-mode and multi-scale feature extraction kernels.
Specifically, each pixel point in the gray-scale map is subjected to traversal calculation, and the pixel points are used for calculation
Taking the local image area matrix with the same size as the characteristic extraction kernel as the center through sampling
Matrix multiplication or matrix subtraction is carried out;
the feature extraction kernel is a two-dimensional matrix f with the size of m × mm*m；
Wherein, the matrix multiplication operation is the multiplication of the two-dimensional matrix of the feature extraction kernel and the local image area matrix, as the following formula (1),
wherein,
the eigenvalues in the feature extraction kernel, i.e., the two-dimensional matrix elements, are represented.
The matrix subtraction is to subtract a local image area matrix from a feature extraction kernel two-dimensional matrix, as shown in the following formula (2),
wherein,
to the pixel pointThe local image area matrix obtained for center sampling
And the feature extraction kernel two-dimensional matrix fm*mThe result after multiplication or subtraction.
It should be noted that the multi-scale, i.e., the size m of the feature extraction kernel, includes a plurality of values, and the multi-scale, i.e., the feature values in the feature extraction kernel
The definition of (1) includes various.
In a specific embodiment, the size m of the feature extraction kernel includes 3, 5, and 9; feature values in a feature extraction kernel
Representing feature values of n-th row and n-th column in a feature extraction kernel, e.g. in a feature kernel
In
The specific feature extraction kernel obtained is as follows,
wherein,
s is an identity matrix of size m.
And generating the feature vector of the pixel point according to the multi-modal feature value definition and different values of the multi-scale feature extraction kernel
As shown in the following formula,
s3, dividing the pixel points into two first clusters according to the result of the feature vector clustering, filtering the clusters to generate a second cluster, and if the second cluster is generated and the number of the pixel points in the second cluster is greater than a first threshold value, possibly having defects; if a second cluster is not generated and the difference value of the pixel points between the two first clusters is greater than a second threshold value, a defect may exist; and under the condition that defects possibly exist, judging whether the coordinate distribution of the pixel point of one first cluster or the second cluster with smaller pixel points in the first cluster is discrete, if so, judging that the first cluster or the second cluster is not defective, otherwise, judging that the first cluster or the second cluster is defective.
Specifically, the S3 includes:
s31, selecting one pixel point as a clustering initial center, traversing all other pixel points of the image, and calculating the Euclidean distance between the clustering initial center and the feature vectors of all other pixel points;
s32, screening out two pixel points with the maximum Euclidean distance from the characteristic vector of the clustering initial center as second clustering centers a and b;
s33, respectively calculating characteristic vector Euclidean distances between each pixel point and the second clustering center a and the second clustering center B, and classifying all the pixel points to obtain a first cluster A of the second clustering center a and a first cluster B of the second clustering center B;
it should be noted that the classification method of the pixel point is specifically that, when the euclidean distance between the pixel point and the feature vector of the second clustering center a is greater than the euclidean distance between the pixel point and the feature vector of the second clustering center B, the pixel point is classified as the first cluster a, otherwise, the pixel point is classified as the first cluster B.
S34, filtering the first cluster A and the first cluster B to generate a second cluster C, and judging whether defects exist or not according to whether the second cluster C is generated or not and the number of pixel points contained in the second cluster C;
the filtering process here is specifically to calculate the distance between all the pixel points in the first cluster a and the second cluster center aThe Euclidean distance of the characteristic vector is calculated and obtained, and the mean value d of the Euclidean distance of the characteristic vector is calculated and obtainedAThe characteristic vector Euclidean distance between the first cluster center a and the second cluster center a is larger than j x dAThe pixel points of (a) are deleted from the first cluster a and added to the second cluster C, where j is a constant term.
In a specific embodiment, the value of j is 2, it should be noted that the value of the constant term j can be adjusted according to different liquid crystal screen images to be measured, and in multiple experiments, it is found that the value of j is about 2, which is the most effective section, and the value of j is 2 as the most preferable section.
The specific step of judging whether a defect may exist according to whether the second cluster C is generated or not and the number of pixels included in the second cluster C is as follows:
when a second cluster C is generated and the number of pixel points in the second cluster C is more than 5, judging that a defect possibly exists; it should be noted that, the value 5 may be changed according to different liquid crystal screen images to be measured.
When the second cluster C is not generated, whether the difference of the pixel point numbers of the first cluster A and the first cluster B is larger than 1/3 of the total pixel point number of the image is judged, and if the difference is larger than 1/3, the defect is judged to exist.
And S35, according to the judgment result of the S34, judging whether the defect exists or not by combining the coordinate distribution of the pixel points in the first cluster A, the first cluster B and the second cluster C.
Specifically, S35 is:
under the condition that a second cluster C appears and the number of pixel points in the second cluster C is more than 5, calculating whether the coordinate distribution of the pixel points in the second cluster C is discrete or not; under the condition that the second cluster C does not appear, selecting a cluster with a small number of pixel points in the first cluster A and the first cluster B, defining the cluster as a cluster Q, and calculating whether the pixel coordinate distribution of the pixel points in the cluster Q is discrete or not; and if the pixel point coordinates are distributed in a discrete manner, judging that no defect exists, and if the pixel point coordinates are distributed in an independently polymerizable region, judging that the defect occurs.
Specifically, whether the coordinate distribution of the pixel points is discrete or not is determined by setting the coordinate distribution of the pixel pointsA distance threshold value d, the distance threshold value
Wherein IwRepresenting the pixel width, I, of an imagehIndicating that the image pixel is high; sequentially traversing the pixel points in the cluster Q or the second cluster C, and calculating the coordinate distribution distance value d of every two pixel points in the cluster Q or the second cluster CPIf d isPD is less than or equal to d, the two pixel points are the coordinate cluster set of the similar pixel points, and the total number N of the newly clustered coordinate distribution clusters is obtainedd；
Presetting a third threshold value T, wherein T is k NtWhere k is a constant term, NtThe total number of pixel points of the cluster Q or the second cluster C is NdWhen the number of the pixels in the cluster Q is larger than or equal to T, the pixel point coordinate distribution of the second cluster C is considered to be discrete, and the pixel point coordinate distribution is judged to be free of defects; otherwise, the coordinate distribution is considered to be in the independently polymerizable region, and the defect is judged.
It should be noted that the value of the constant term k may also vary according to different liquid crystal screen images to be measured, and a large number of experiments show that the preferred value range is 0.15-0.35.
In a second aspect, an electronic device is provided that includes a memory and a processor;
the memory stores computer-executable instructions;
the processor executes the computer-executable instructions stored in the memory, so that the processor executes the liquid crystal display defect detection method for establishing the cluster based on the single-pixel characteristic clustering.
Compared with the prior art, the technical scheme of the invention can bring the following improvement effects,
(1) compared with the existing deep learning technical scheme based on small samples, the method does not need to collect the defect samples, is more practical in algorithm floor application, and effectively improves the efficiency of the defect detection algorithm due to small scale of parameters;
(2) compared with the traditional algorithms such as local pixel values and the like, the defect detection is more comprehensive, and can be used for detecting the zigzag defects and large-area defects of the liquid crystal display.
The above is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes will occur to those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A liquid crystal screen defect detection method based on single pixel characteristic clustering cluster building cluster is disclosed, firstly, RGB three-channel image of liquid crystal screen is obtained, which is characterized by also comprising the following steps:
s1, obtaining gray values and r, g and b channel values of all pixel points of the image, and converting the image into a gray map;
s2, performing traversal calculation on each pixel point in the gray-scale image, and generating a feature vector of the pixel point through a multi-mode and multi-scale feature extraction kernel;
s3, dividing the pixel points into two first clusters according to the result of the feature vector clustering, filtering the clusters to generate a second cluster, and if the second cluster is generated and the number of the pixel points in the second cluster is greater than a first threshold value, possibly having defects; if a second cluster is not generated and the difference value of the pixel points between the two first clusters is greater than a second threshold value, a defect may exist; and under the condition that defects possibly exist, judging whether the coordinate distribution of the pixel point of one first cluster or the second cluster with smaller pixel points in the first cluster is discrete, if so, judging that the first cluster or the second cluster is not defective, otherwise, judging that the first cluster or the second cluster is defective.
2. The method for detecting defects of a liquid crystal display based on single-pixel feature cluster building cluster according to claim 1, wherein the traversing calculation of each pixel point in the gray-scale map specifically comprises:
by pixel points
Taking the local image area matrix with the same size as the characteristic extraction kernel as the center through sampling
Then carrying out matrix multiplication or matrix subtraction operation;
the feature extraction kernel is a two-dimensional matrix f with the size of m × mm*m；
Wherein, the matrix multiplication operation is the multiplication of the two-dimensional matrix of the feature extraction kernel and the local image area matrix, as the following formula (1),
the matrix subtraction is to subtract a local image area matrix from a feature extraction kernel two-dimensional matrix, as shown in the following formula (2),
wherein,
to the pixel point
The local image area matrix obtained for center sampling
And the feature extraction kernel two-dimensional matrix fm*mThe result after multiplication or subtraction.
3. The method for detecting the defects of the liquid crystal display based on the single-pixel feature cluster building cluster as claimed in claim 2, wherein the feature vectors for generating the pixel points through the multi-modal and multi-scale feature extraction kernel are specifically as follows:
the multi-scale value, namely the size m of the feature extraction kernel, comprises a plurality of values, and the multi-scale value is the feature value in the feature extraction kernel
The definition of (a) includes a plurality;
and generating the feature vector of the pixel point according to the multi-modal feature value definition and different values of the multi-scale feature extraction kernel
4. The method for detecting defects of a liquid crystal display based on single-pixel feature cluster building cluster according to claim 1, wherein the S3 specifically includes:
s31, selecting one pixel point as a clustering initial center, traversing all other pixel points of the image, and calculating the Euclidean distance between the clustering initial center and the feature vectors of all other pixel points;
s32, screening out two pixel points with the maximum Euclidean distance from the characteristic vector of the clustering initial center as second clustering centers a and b;
s33, respectively calculating characteristic vector Euclidean distances between each pixel point and the second clustering center a and the second clustering center B, and classifying all the pixel points to obtain a first cluster A of the second clustering center a and a first cluster B of the second clustering center B;
s34, filtering the first cluster A and the first cluster B to generate a second cluster C, and judging whether defects exist or not according to whether the second cluster C is generated or not and the number of pixel points contained in the second cluster C;
and S35, according to the judgment result of the S34, judging whether the defect exists or not by combining the coordinate distribution of the pixel points in the first cluster A, the first cluster B and the second cluster C.
5. The method for detecting defects of a liquid crystal display based on single-pixel feature cluster building cluster according to claim 4, wherein the classifying all the pixel points in the step S33 is specifically as follows:
and when the Euclidean distance between the pixel point and the characteristic vector of the second clustering center a is greater than that between the pixel point and the characteristic vector of the second clustering center B, classifying the pixel point as a first cluster A, otherwise, classifying the pixel point as a first cluster B.
6. The method for detecting defects of a liquid crystal display based on single-pixel feature cluster building cluster according to claim 5, wherein the filtering process in S34 specifically comprises:
calculating characteristic vector Euclidean distances between all pixel points in the first cluster A and the second cluster center a, and calculating and obtaining a characteristic vector Euclidean distance mean value dAThe characteristic vector Euclidean distance between the first cluster center a and the second cluster center a is larger than j x dAThe pixel points of (a) are deleted from the first cluster a and added to the second cluster C, where j is a constant term.
7. The method for detecting the defects of the liquid crystal display based on the single-pixel feature cluster established cluster is characterized in that the step of judging whether the defects possibly exist according to whether the second cluster C is generated or not and the number of pixels contained in the second cluster C is specifically as follows:
when a second cluster C is generated and the number of pixel points in the second cluster C is more than 5, judging that a defect possibly exists;
when the second cluster C is not generated, whether the difference of the pixel point numbers of the first cluster A and the first cluster B is larger than 1/3 of the total pixel point number of the image is judged, and if the difference is larger than 1/3, the defect is judged to exist.
8. The method for detecting defects of a liquid crystal display based on single-pixel feature cluster building cluster according to claim 7, wherein the step S35 specifically comprises:
under the condition that a second cluster C appears and the number of pixel points in the second cluster C is more than 5, calculating whether the coordinate distribution of the pixel points in the second cluster C is discrete or not; under the condition that the second cluster C does not appear, selecting a cluster with a small number of pixel points in the first cluster A and the first cluster B, defining the cluster as a cluster Q, and calculating whether the pixel coordinate distribution of the pixel points in the cluster Q is discrete or not; and if the pixel point coordinates are distributed in a discrete manner, judging that no defect exists, and if the pixel point coordinates are distributed in an independently polymerizable region, judging that the defect occurs.
9. The method for detecting the defects of the liquid crystal display based on the single-pixel feature cluster building cluster of claim 8, wherein whether the coordinate distribution of the pixel points is discrete is specifically as follows:
setting a pixel point coordinate distribution distance threshold d, wherein the distance threshold
Wherein IwRepresenting the pixel width, I, of an imagehIndicating that the image pixel is high; sequentially traversing the pixel points in the cluster Q or the second cluster C, and calculating the coordinate distribution distance value d of every two pixel points in the cluster Q or the second cluster CPIf d isPD is less than or equal to d, the two pixel points are the coordinate cluster set of the similar pixel points, and the total number N of the newly clustered coordinate distribution clusters is obtainedd；
Presetting a third threshold value T, wherein T is k NtWhere k is a constant term, NtThe total number of pixel points of the cluster Q or the second cluster C is NdWhen the number of the pixels in the cluster Q is larger than or equal to T, the pixel point coordinate distribution of the second cluster C is considered to be discrete, and the pixel point coordinate distribution is judged to be free of defects; otherwise, the coordinate distribution is considered to be in the independently polymerizable region, and the defect is judged.
10. An electronic device comprising a memory and a processor;
the memory stores computer-executable instructions;
the processor executes the computer-executable instructions stored in the memory, so that the processor executes the liquid crystal display defect detection method based on clustering of single-pixel features according to any one of claims 1 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111204952.1A CN114022415B (en) | 2021-10-15 | 2021-10-15 | Liquid crystal display defect detection method based on single-pixel feature clustering cluster establishment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111204952.1A CN114022415B (en) | 2021-10-15 | 2021-10-15 | Liquid crystal display defect detection method based on single-pixel feature clustering cluster establishment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114022415A true CN114022415A (en) | 2022-02-08 |
| CN114022415B CN114022415B (en) | 2022-06-28 |
Family
ID=80056283
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111204952.1A Active CN114022415B (en) | 2021-10-15 | 2021-10-15 | Liquid crystal display defect detection method based on single-pixel feature clustering cluster establishment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114022415B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115082722A (en) * | 2022-08-22 | 2022-09-20 | 四川金信石信息技术有限公司 | Equipment defect detection method, system, terminal and medium based on forward sample |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104978748A (en) * | 2015-07-06 | 2015-10-14 | 电子科技大学 | Liquid crystal screen defect detection method based on local pixel values |
| CN110009607A (en) * | 2019-03-25 | 2019-07-12 | 深圳市洲明科技股份有限公司 | Show screen dead pixel detection method, device, computer equipment and storage medium |
| CN110021023A (en) * | 2019-03-05 | 2019-07-16 | 西安工程大学 | A kind of electronics cloth defect segmentation method |
| US20190340739A1 (en) * | 2018-05-02 | 2019-11-07 | Samsung Display Co., Ltd. | Defect detection apparatus and method |
| CN110954553A (en) * | 2019-10-10 | 2020-04-03 | 段美华 | Lens optical detection system |
| CN111340752A (en) * | 2019-12-04 | 2020-06-26 | 京东方科技集团股份有限公司 | Screen detection method and device, electronic equipment and computer readable storage medium |
| CN111612757A (en) * | 2020-05-18 | 2020-09-01 | 苏州精濑光电有限公司 | Screen crack detection method, device, equipment and storage medium |
| CN111815630A (en) * | 2020-08-28 | 2020-10-23 | 歌尔股份有限公司 | Defect detection method and device for LCD screen |
| CN112419229A (en) * | 2020-10-15 | 2021-02-26 | 合肥联宝信息技术有限公司 | Display screen linear defect detection method and device and storage medium |
-
2021
- 2021-10-15 CN CN202111204952.1A patent/CN114022415B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104978748A (en) * | 2015-07-06 | 2015-10-14 | 电子科技大学 | Liquid crystal screen defect detection method based on local pixel values |
| US20190340739A1 (en) * | 2018-05-02 | 2019-11-07 | Samsung Display Co., Ltd. | Defect detection apparatus and method |
| CN110021023A (en) * | 2019-03-05 | 2019-07-16 | 西安工程大学 | A kind of electronics cloth defect segmentation method |
| CN110009607A (en) * | 2019-03-25 | 2019-07-12 | 深圳市洲明科技股份有限公司 | Show screen dead pixel detection method, device, computer equipment and storage medium |
| CN110954553A (en) * | 2019-10-10 | 2020-04-03 | 段美华 | Lens optical detection system |
| CN111340752A (en) * | 2019-12-04 | 2020-06-26 | 京东方科技集团股份有限公司 | Screen detection method and device, electronic equipment and computer readable storage medium |
| CN111612757A (en) * | 2020-05-18 | 2020-09-01 | 苏州精濑光电有限公司 | Screen crack detection method, device, equipment and storage medium |
| CN111815630A (en) * | 2020-08-28 | 2020-10-23 | 歌尔股份有限公司 | Defect detection method and device for LCD screen |
| CN112419229A (en) * | 2020-10-15 | 2021-02-26 | 合肥联宝信息技术有限公司 | Display screen linear defect detection method and device and storage medium |
Non-Patent Citations (4)
| Title |
|---|
| GHISLAIN TAKAM TCHENDJOU等: "Detection, Location and Concealment of Defective Pixels in Image Sensors", 《IEEE TRANSACTIONS ON EMERGING TOPICS IN COMPUTING》 * |
| 夏晓云: "基于MapReduce的液晶屏缺陷检测技术研究", 《中国优秀硕士学位论文全文数据库_信息科技辑》 * |
| 汪志亮等: "面向OLED屏像素缺陷检测的新方法", 《计算机工程与应用》 * |
| 覃晓伟: "基于机器视觉的手机屏幕坏点检测系统研究", 《中国优秀硕士学位论文全文数据库_信息科技辑》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115082722A (en) * | 2022-08-22 | 2022-09-20 | 四川金信石信息技术有限公司 | Equipment defect detection method, system, terminal and medium based on forward sample |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114022415B (en) | 2022-06-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106650770B (en) | Mura defect detection method based on sample learning and human eye visual characteristics | |
| US10860879B2 (en) | Deep convolutional neural networks for crack detection from image data | |
| CN107845087A (en) | The detection method and system of the uneven defect of liquid crystal panel lightness | |
| CN110599552B (en) | pH test paper detection method based on computer vision | |
| CN103761529B (en) | A kind of naked light detection method and system based on multicolour model and rectangular characteristic | |
| Zhang et al. | Separation of touching grain kernels in an image by ellipse fitting algorithm | |
| CN112862744B (en) | Intelligent detection method for internal defects of capacitor based on ultrasonic image | |
| CN111640104B (en) | Visual detection method for screw assembly | |
| CN108520514B (en) | Consistency detection method for electronic elements of printed circuit board based on computer vision | |
| CN111047655A (en) | High-definition camera cloth defect detection method based on convolutional neural network | |
| US20200096454A1 (en) | Defect detection system for aircraft component and defect detection method for aircraft component | |
| CN104616275A (en) | Defect detecting method and defect detecting device | |
| CN112819844B (en) | Image edge detection method and device | |
| CN110648330B (en) | Defect detection method for camera glass | |
| CN111368756A (en) | Visible light-based method and system for quickly identifying open fire smoke | |
| KR20170127269A (en) | Method and apparatus for detecting and classifying surface defect of image | |
| CN114022415B (en) | Liquid crystal display defect detection method based on single-pixel feature clustering cluster establishment | |
| CN111833347A (en) | Transmission line damper defect detection method and related device | |
| CN110349133B (en) | Object surface defect detection method and device | |
| Kuo et al. | Automated inspection of micro-defect recognition system for color filter | |
| CN113221812A (en) | Training method of face key point detection model and face key point detection method | |
| CN108830834B (en) | Automatic extraction method for video defect information of cable climbing robot | |
| CN113470015B (en) | Water body shaking detection and analysis method and system based on image processing | |
| CN113870255B (en) | Mini LED product defect detection method and related equipment | |
| CN114705696A (en) | Defect detection method based on digital image processing and automatic optical detection |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |