CN116596922B - Production quality detection method of solar water heater - Google Patents
Production quality detection method of solar water heater Download PDFInfo
- Publication number
- CN116596922B CN116596922B CN202310868108.1A CN202310868108A CN116596922B CN 116596922 B CN116596922 B CN 116596922B CN 202310868108 A CN202310868108 A CN 202310868108A CN 116596922 B CN116596922 B CN 116596922B
- Authority
- CN
- China
- Prior art keywords
- texture
- pixel point
- foreground
- determining
- responsivity
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 title claims abstract description 21
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 18
- 238000001514 detection method Methods 0.000 title abstract description 11
- 239000013598 vector Substances 0.000 claims abstract description 103
- 230000007547 defect Effects 0.000 claims abstract description 45
- 230000011218 segmentation Effects 0.000 claims abstract description 30
- 238000000034 method Methods 0.000 claims abstract description 27
- 239000011159 matrix material Substances 0.000 claims description 55
- 230000002950 deficient Effects 0.000 claims description 12
- 238000000513 principal component analysis Methods 0.000 claims description 9
- 230000004043 responsiveness Effects 0.000 claims description 9
- 238000006243 chemical reaction Methods 0.000 claims description 8
- 238000003708 edge detection Methods 0.000 claims description 4
- 238000000354 decomposition reaction Methods 0.000 claims description 3
- 230000007812 deficiency Effects 0.000 claims description 2
- 230000001131 transforming effect Effects 0.000 abstract 1
- 230000004044 response Effects 0.000 description 7
- 230000009466 transformation Effects 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 3
- 238000010606 normalization Methods 0.000 description 3
- 241000287196 Asthenes Species 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 2
- 230000000295 complement effect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 239000002184 metal Substances 0.000 description 2
- 229910044991 metal oxide Inorganic materials 0.000 description 2
- 150000004706 metal oxides Chemical class 0.000 description 2
- 239000004065 semiconductor Substances 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000003466 welding Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-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
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/40—Analysis of texture
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E10/00—Energy generation through renewable energy sources
- Y02E10/40—Solar thermal energy, e.g. solar towers
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method for detecting production quality of a solar water heater, which belongs to the technical field of image processing and comprises the following steps: transforming the acquired image to obtain a subband; obtaining a subband main feature vector to determine a texture detail result and a foreground pixel point, and calculating texture responsivity and texture missing degree of the foreground pixel point; and determining a segmentation threshold value for global threshold segmentation based on the texture defect degree, calculating the ratio of the area of the defect region of the segmented defect region to the area of the foreground image, and determining the heat collecting plate with the ratio larger than a standard value as unqualified. Therefore, based on the sub-band for obtaining the solar heat collection plate image, texture detail processing results are determined based on main features of the sub-band to obtain foreground pixel points, texture responsivity and texture missing degree of the foreground pixel points are obtained, a defect part is segmented from the foreground image based on the texture missing degree, and finally a production quality detection result is determined based on the area ratio of the defect part, so that the accuracy of quality detection of the solar water heater is greatly improved.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a production quality detection method of a solar water heater.
Background
Because the solar heat collecting plate is made of a plurality of metal sheets, scribing treatment is needed before welding, and the influence of the surface grains of the metal sheets is added, the defects of the plate existing in the deep color of the grains are not easy to detect by a common defect detection method. Therefore, the detection of the production quality of the solar water heater often has the problem of low accuracy.
Disclosure of Invention
The invention provides a method for detecting the production quality of a solar water heater, which aims to improve the accuracy of the detection of the production quality of the solar water heater.
In order to achieve the above object, the present invention provides a method for detecting production quality of a solar water heater, the method comprising:
performing gray level conversion and wavelet conversion on the acquired solar heat collecting plate image to obtain four sub-bands, wherein each sub-band comprises a low-frequency sub-band and three high-frequency sub-bands;
acquiring a main feature vector of the sub-band, and determining texture detail results of each pixel point based on the main feature vector;
determining foreground pixel points based on texture detail processing results of the pixel points, and calculating texture responsivity of the foreground pixel points;
determining texture missing degree of the foreground pixel point based on the minimum distance value of the foreground pixel point and the edge pixel point and the texture responsivity;
determining a segmentation threshold value based on the texture defect degree of each foreground pixel point, and performing global threshold segmentation on a foreground image corresponding to the foreground pixel point based on the segmentation threshold value to obtain a defect region;
and calculating the ratio of the area of the defect area to the area of the foreground image, and determining the heat collecting plate corresponding to the foreground image with the ratio larger than the standard value as the unqualified solar heat collecting plate.
Optionally, the obtaining the main feature vector of the subband, and determining the texture detail result of each pixel point based on the main feature vector includes:
obtaining a vector matrix of the sub-bands, obtaining a first covariance matrix based on the vector matrix, and determining a main feature vector of each sub-band based on a feature value matrix and a feature vector matrix of the first covariance matrix;
and obtaining a main feature vector of the pixel point at the specified position in the coordinate system, and obtaining a texture detail result of the corresponding pixel point based on the main feature vector.
Optionally, the obtaining the vector matrix of the subband, obtaining a first covariance matrix based on the vector matrix, and determining the principal eigenvectors of each subband based on the eigenvalue matrix and the eigenvector matrix of the first covariance matrix includes:
normalizing the pixel value of each pixel point in the sub-band to obtain a normalized pixel value;
constructing a vector matrix of subbands based on the normalized pixel values;
determining vectors in the same direction in the vector matrix as characteristic vectors of subbands, wherein the characteristic vectors are arranged into a characteristic vector matrix according to columns;
compressing eigenvectors of the sub-bands into one based on principal component analysis, and constructing a first covariance matrix of the eigenvector matrix;
performing eigenvalue decomposition on the first covariance matrix to obtain a first eigenvalue and a first eigenvector;
and determining the first eigenvector corresponding to the maximum value in the first eigenvectors as a main eigenvector.
Optionally, the obtaining the main feature vector of the pixel point at the specified position in the coordinate system, and obtaining the texture detail result of the corresponding pixel point based on the main feature vector includes:
constructing a coordinate system by taking the upper left corner of a subband as an origin, taking the horizontal direction as an x axis and taking the vertical direction as a y axis, and obtaining a main feature vector of a pixel point at a specified position in the coordinate system;
multiplying the principal eigenvector of the pixel point by the transpose of the principal component vector to obtain a high-dimensional vector of the pixel point after inversion;
and normalizing the high-dimensional vector to obtain a texture detail result of the pixel point based on the main feature vector.
Optionally, the determining the foreground pixel point based on the texture detail processing result of each pixel point, and calculating the texture responsivity of each foreground pixel point includes:
calculating the maximum value of texture detail results of three high-frequency feature vectors of the pixel points, comparing the texture detail results of the low-frequency feature vectors with the maximum value of the texture detail results, and determining the corresponding pixel point as a background pixel point if the texture detail results of the low-frequency feature vectors are larger than the maximum value of the texture detail results;
if the texture detail result of the low-frequency feature vector is smaller than or equal to the maximum value of the texture detail result, determining the corresponding pixel point as a foreground pixel point;
calculating texture responsivity of the foreground pixel points based on texture detail results;
and determining the texture responsivity of the foreground pixel point according to the texture responsivity of the texture detail result.
Optionally, before determining the texture missing degree of the foreground pixel point based on the minimum distance value between the foreground pixel point and the edge pixel point and the texture responsivity, the method further includes:
performing edge detection on the gray level image through a canny operator to obtain an edge image;
respectively establishing a search window with preset size on the gray level image by taking each pixel point as a center, and gradually increasing the size of the search window until edge pixel points appear in the search window;
and calculating the minimum distance value between each pixel point and the corresponding edge pixel point.
Optionally, the determining the texture defect degree of the foreground pixel point based on the minimum distance value between the foreground pixel point and the edge pixel point and the texture responsivity includes:
calculating the average texture responsivity and the standard deviation of the texture responsivity of each foreground pixel point;
determining the weight of each foreground pixel point based on the average texture responsivity, the texture responsivity standard deviation and the texture missing degree of the foreground pixel point;
and determining the product of the texture missing degree, the minimum distance value and the weight of the foreground pixel point as the texture missing degree of the corresponding foreground pixel point.
Optionally, determining a segmentation threshold based on the texture defect degree of each foreground pixel point, and performing global threshold segmentation on the foreground image corresponding to the foreground pixel point based on the segmentation threshold, to obtain the defect region includes:
calculating the edge texture defect degree average value of all the edge pixel points;
taking the Euclidean distance between the foreground pixel point texture missing degree and the edge texture missing degree as a segmentation threshold value, and segmenting the foreground image based on the segmentation threshold value;
and determining the foreground pixel point with the Euclidean distance larger than the experience value as a defective pixel point, and determining the region corresponding to the defective pixel point as a defective region.
Compared with the prior art, the production quality detection method of the solar water heater provided by the invention has the advantages that the acquired solar heat collecting plate image is subjected to gray level conversion and wavelet transformation to obtain four sub-bands, and the sub-bands comprise a low-frequency sub-band and three high-frequency sub-bands; acquiring a main feature vector of the sub-band, and determining texture detail results of each pixel point based on the main feature vector; determining foreground pixel points based on texture detail processing results of the pixel points, and calculating texture responsivity of the foreground pixel points; determining texture missing degree of the foreground pixel point based on the minimum distance value of the foreground pixel point and the edge pixel point and the texture responsivity; determining a segmentation threshold value based on the texture defect degree of each foreground pixel point, and performing global threshold segmentation on a foreground image corresponding to the foreground pixel point based on the segmentation threshold value to obtain a defect region; and calculating the ratio of the area of the defect area to the area of the foreground image, and determining the heat collecting plate corresponding to the foreground image with the ratio larger than the standard value as the unqualified solar heat collecting plate. In this way, the low-frequency sub-band and the high-frequency sub-band are obtained based on the transformation of the solar heat collecting plate image, then the texture detail processing result is determined based on the main characteristics of the sub-bands to obtain the foreground pixel point, the texture responsivity and the texture missing degree of the foreground pixel point are further obtained, the defect part is segmented from the foreground image based on the texture missing degree, and finally the production quality result of the heat collecting plate of the solar water heater is determined based on the area ratio of the defect part, so that the accuracy of quality detection of the solar water heater is greatly improved.
Drawings
FIG. 1 is a schematic flow chart of an embodiment of a method for detecting production quality of a solar water heater according to the present invention;
FIG. 2 is a schematic diagram of a refinement flow chart of an embodiment of a method for detecting production quality of a solar water heater.
The achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Referring to fig. 1, fig. 1 is a schematic flow chart of a first embodiment of a method for detecting production quality of a solar water heater according to the present invention.
As shown in fig. 1, a first embodiment of the present invention provides a method for detecting production quality of a solar water heater, the method comprising:
step S101, gray level conversion and wavelet transformation are carried out on an acquired solar heat collecting plate image to obtain four sub-bands, wherein each sub-band comprises a low-frequency sub-band and three high-frequency sub-bands;
the main purpose of this embodiment is to detect fine defects on the heat collecting plate of a solar water heater. Firstly, performing image acquisition on the welded solar heat collecting plate through a complementary metal oxide semiconductor (Complementary Metal Oxide Semiconductor, CMOS) camera to obtain an RGB image of the solar heat collecting plate, and performing gray level conversion on the RGB image of the solar heat collecting plate to obtain a gray level image. Then, a Haar wavelet basis is selected to carry out wavelet transformation on the gray level image, so as to obtain a low-frequency sub-band and three high-frequency sub-bands, wherein the low-frequency sub-band is denoted as LL, and the high-frequency sub-bands are respectively denoted as LH, HL and HH. The low frequency sub-band represents the basic information of the image, the high frequency sub-band contains the edge and texture information in the image, the high frequency sub-band is subjected to threshold processing to remove noise, then the low frequency sub-band is combined with the low frequency sub-band to obtain a denoised image, and histogram equalization is carried out on the denoised image to realize image enhancement. All the steps are known in the prior art, and an implementer can carry out algorithm replacement according to the function implementation.
Step S102, obtaining a main feature vector of the sub-band, and determining texture detail results of each pixel point based on the main feature vector;
referring to fig. 2, fig. 2 is a schematic diagram of a refinement flow chart of an embodiment of a method for detecting production quality of a solar water heater according to the present invention, as shown in fig. 2, step S102 includes:
step S1021, obtaining a vector matrix of the sub-bands, obtaining a first covariance matrix based on the vector matrix, and determining a main eigenvector of each sub-band based on an eigenvalue matrix and an eigenvector matrix of the first covariance matrix;
specifically, the pixel values of the pixel points in the sub-band are normalized to obtain normalized pixel values；
The gray level image corresponding to the solar heat collecting plate image is subjected to wavelet transformation to obtain low-frequency characteristics when the image is denoisedAnd high frequency characteristics->、/>And->Four subbands, subband vectorization using Principal Component Analysis (PCA), setThe size of the sub-band is +.>Is marked as->The size of the remaining sub-bands is also +.>. Constructing a coordinate system by taking the upper left corner of the sub-band as an origin, taking the horizontal direction as the x axis and the vertical direction as the y axis, and constructing pixels +.>For the values of the ith row and jth column pixels in the subband, a matrix is constructed with the pixel values of all pixels in the subband, for +.>Normalization is performed, and the normalized pixel value is expressed asThen:
；
wherein,,for the pixel value before normalization, +.>And->The mean value and standard deviation of the pixel values of all the pixel points are respectively. Normalized +.>The dimensional influence of element values in the matrix can be eliminated, so that comparison can be performed among different pixels, and the reliability and the interpretability of data are improved.
Constructing a vector matrix of subbands based on the normalized pixel values: all pixels in the original matrix are normalized and listed as new sizeMatrix of->。
Determining vectors in the same direction in the vector matrix as characteristic vectors of subbands, wherein the characteristic vectors are arranged into a characteristic vector matrix according to columns; the normalized vector in the same direction is the characteristic vector of the sub-band, the characteristic vector is ordered according to the column, and the characteristic vector is provided withIs (are) and->。
Compressing eigenvectors of the sub-bands into one based on principal component analysis, and constructing a first covariance matrix of the eigenvector matrix; will beThe feature vectors form a ++>Feature vector matrix>Feature vectors of subbands are compressed to one using Principal Component Analysis (PCA): constructing a feature vector matrix>Is>First covariance matrix->Is +.>。
Performing eigenvalue decomposition on the first covariance matrix to obtain a first eigenvalue and a first eigenvector; a first feature corresponding to the maximum value in the first feature vectorThe vector is determined as the principal eigenvector. For the first covariance matrixDecomposing the characteristic value to obtain a first characteristic value +.>And a first feature vector->Taking the feature vector with the largest first feature valueAs a compressed low frequency feature vector. Similarly, the method is used for calculating the high-frequency characteristic vectors of the other three high-frequency characteristic sub-bands>，/>，/>. The present embodiment refers to the low-frequency feature vector and the high-frequency feature vector collectively as a main feature vector.
Step S1022, obtaining a main feature vector of the pixel point at the specified position in the coordinate system, and obtaining a texture detail result of the corresponding pixel point based on the main feature vector.
Constructing a coordinate system by taking the upper left corner of a subband as an origin, taking the horizontal direction as an x axis and taking the vertical direction as a y axis, and obtaining a main feature vector of a pixel point at a specified position in the coordinate system;
after PCA compression, the main eigenvectors of the four subbands are obtained: a low frequency eigenvector for the low frequency subband and a high frequency eigenvector for the high frequency subband. By high-frequency eigenvectors in the principal eigenvectorFind the coordinate position for example as +.>Is +.>And obtaining the corresponding high-frequency characteristic vector. For a size of +.>The high frequency eigenvector of which can be expressed as a subband of length +.>Is a vector of (a). Thus, a subband is to be obtained>Middle->The high frequency eigenvector corresponding to the position can be calculated +.>And (3) the position in the vector and fetching the corresponding element. The calculation method comprises the following steps:
wherein,,is the width of the subband image, < >>And->Is the coordinates of the pixel in the subband coordinate axes. By calculation ofCan be in the sub-band->Extracting corresponding elements from the high frequency eigenvectors of (2) to obtain +.>High-frequency feature vector corresponding to position->。
Multiplying the principal eigenvector of the pixel point by the transpose of the principal component vector to obtain a high-dimensional vector of the pixel point after inversion;
and normalizing the high-dimensional vector to obtain a texture detail result of the pixel point based on the main feature vector.
For pair willMultiplying the transpose of the principal component vector of PCA to obtain the inverse transformed high-dimensional vector +.>For->Normalization processing is carried out to obtain a texture detail result +.>，/>Is a value which can be regarded as pixel +.>At->Projection representation of a high-frequency feature space, which measures +.>The variation component and amplitude in the direction of the high frequency features.
Repeating the above calculation to obtain each pixel in the image based on the respective principal eigenvector，/>，/>Texture detail results->，/>，/>。
Texture detail results for high frequency feature vectors for each pixel，/>，/>Representing the degree of response of each pixel to the corresponding high-frequency feature, the higher the value, the more intense the response, and the texture detail result of the low-frequency feature vector of each pixel is similarly +.>Representing the degree of response of each pixel to the low frequency characteristic, the higher the value, the more aggressive the response.
The original threshold segmentation is to segment the gray value of each pixel of the image as input, and the image can be divided into two parts according to the distribution condition of the gray value of the image. The global thresholding method is improved with this result based on the texture detail result of the principal eigenvector of each pixel that has been obtained.
Step S103, determining foreground pixel points based on texture detail processing results of the pixel points, and calculating texture responsivity of the foreground pixel points;
firstly, calculating the maximum value of texture detail results of three high-frequency feature vectors of a pixel pointTexture detail result of low frequency feature vector +.>Maximum value of texture detail result +.>Comparing, if the texture detail result of the low-frequency feature vector is larger than the maximum value of the texture detail result, determining the corresponding pixel point as a background pixel point; namely->The low frequency response of the pixel is larger, and the pixel is a background pixel with high probability.
If the texture detail result of the low-frequency feature vector is smaller than or equal to the maximum value of the texture detail result, determining the corresponding pixel point as a foreground pixel point;indicating that the pixel has a small low frequency response.
Calculating texture responsivity of the foreground pixel point based on the texture detail result;
will be based on texture detail resultsThe texture responsiveness of (a) is expressed as +.>Then:
representing the degree of resonance of the pixel for the two principal eigenvectors,introduces->Phase value of>The method represents the sum of the amplitudes of the two principal eigenvectors, and the texture responsivity of the foreground pixel point can be well represented by using the amplitude, the phase and the resonance degree of the texture details of the principal eigenvectors. Thereafter computing texture detail based results using the same methodAnd->Texture responsivity of (2), expressed as +.>And->。
Finally, determining the texture responsivity of the foreground pixel point according to the texture responsivity of the texture detail result;
wherein n is 1,2,3.
When (when)When the detail information representing the foreground pixel point is larger, the detail information is +.>The magnitude of the sensitivity of the foreground pixel point to image details is represented, and the texture responsiveness is more obvious for the defective pixel response in a small scale range.
Step S104, determining the texture missing degree of the foreground pixel point based on the minimum distance value between the foreground pixel point and the edge pixel point and the texture responsivity;
the present embodiment further needs to determine a minimum distance value between a pixel point and an edge pixel point before calculating the texture missing degree, specifically:
performing edge detection on the gray level image through a canny operator to obtain an edge image; the present embodiment samples the well-known technique of the canny operator for edge detection. And identifying corresponding edge pixel points after obtaining the edge image.
Respectively establishing a search window with preset size on the gray level image by taking each pixel point as a center, and gradually increasing the size of the search window until edge pixel points appear in the search window; representing the search window size as. For pixel dot->Creating a pixel by +.>The size of the center is +.>Search window of->From 3, the pixel is gradually increased until there are edge pixels of the edge image in the search window.
And calculating the minimum distance value between each pixel point and the corresponding edge pixel point. In this embodiment, the euclidean distance is used to calculate the euclidean distance between the pixel point and the corresponding edge pixel point, the minimum euclidean distance is determined as the minimum distance value, and the minimum distance value is denoted as D.
After determining the minimum distance value and texture responsivity, further calculating the texture missing degree:
calculating the average texture responsivity of each foreground pixel pointAnd texture responsivity standard deviation->；
Based on the average texture responsivityStandard deviation of texture responsiveness>And determining the weight of each foreground pixel point according to the texture missing degree of the foreground pixel point;
determining the product of the texture missing degree, the minimum distance value and the weight of the foreground pixel point as the texture missing degree of the corresponding foreground pixel point;
representing texture deficiency degree asThen:
texture missing degree of foreground pixel pointThe pixel abnormality degree of the heat collecting plate is represented, and the larger the value is, the stronger the abnormality of the pixel point is>For the texture responsivity of the pixel, +.>For the minimum distance value between each pixel point and the corresponding edge pixel point, when the texture responsivity W and the minimum distance value D become larger, representing the texture responseThe degree becomes stronger or the pixel point is far from the edge, reflecting that the pixel point is most likely to be a defective pixel point, and the corresponding texture defect degree is +.>And also becomes larger, otherwise the texture deletion degree is +.>And consequently decreases. />And->Texture responsivity of all pixels respectively +.>Is defined as the mean value and standard deviation of (c),according to texture responsiveness->Mean value->Is given different weights, the values of which follow the differences in (a)Gradually increasing with increasing texture responsiveness +.>And->The larger the difference, the larger the difference between the texture responsiveness of the pixel point and the average texture responsiveness of the whole image. Reflecting that the texture responsiveness of the pixel point has a great difference from the average texture features in the neighborhood, it may mean that the pixel point has a special texture structure or an abnormal condition. Texture responsiveness near average>The method can give smaller weight, has a smoothing effect on the texture missing degree, and has little influence on the final numerical value.
Step S105, determining a segmentation threshold value based on the texture missing degree of each foreground pixel point, and performing global threshold segmentation on a foreground image corresponding to the foreground pixel point based on the segmentation threshold value to obtain a defect region;
calculating the edge texture defect degree average value of all the edge pixel points, and expressing the edge texture defect degree average value as；
Taking the Euclidean distance between the foreground pixel point texture missing degree and the edge texture missing degree as a segmentation threshold value, and segmenting the foreground image based on the segmentation threshold value; representing Euclidean distance as。
And determining the foreground pixel point with the Euclidean distance larger than the experience value as a defective pixel point, and determining the region corresponding to the defective pixel point as a defective region. The empirical value is set according to actual needs, the present embodiment determines the empirical value to be 20,is a defective pixel.
And S106, calculating the ratio of the area of the defect area to the area of the foreground image of the defect area, and determining the heat collecting plate corresponding to the foreground image with the ratio larger than the standard value as the unqualified solar heat collecting plate.
Because the pixel points of the defect part are in the most detailed texture information, the texture defect degree of the heat collecting plate of the pixel points should belong to the largest part, and the difference value between the texture defect degree of the defect pixel point and the texture defect degree of the edge pixel point is larger, the improved global threshold method divides the whole image into the defect part and the background part, and calculates the area of the defect partArea of foreground image +.>The ratio S/M is set to a standard value +.>When S/M is greater than the standard value +.>And if the corresponding solar heat collecting plate is unqualified, otherwise, the corresponding solar heat collecting plate is qualified in quality. Standard value->The setting can be performed by the user according to actual conditions, and is not unique. The method not only can accurately detect very fine cracks and bubbles on the solar heat collecting plate due to the technical problem, but also can detect large-scale stains and defects, thereby greatly ensuring the product quality.
According to the embodiment, through the scheme, the acquired solar heat collecting plate image is subjected to gray level conversion and wavelet transformation to obtain four sub-bands, wherein the sub-bands comprise a low-frequency sub-band and three high-frequency sub-bands; acquiring a main feature vector of the sub-band, and determining texture detail results of each pixel point based on the main feature vector; determining foreground pixel points based on texture detail processing results of the pixel points, and calculating texture responsivity of the foreground pixel points; determining texture missing degree of the foreground pixel point based on the minimum distance value of the foreground pixel point and the edge pixel point and the texture responsivity; determining a segmentation threshold value based on the texture defect degree of each foreground pixel point, and performing global threshold segmentation on a foreground image corresponding to the foreground pixel point based on the segmentation threshold value to obtain a defect region; and calculating the ratio of the area of the defect area to the area of the foreground image, and determining the heat collecting plate corresponding to the foreground image with the ratio larger than the standard value as the unqualified solar heat collecting plate. In this way, the low-frequency sub-band and the high-frequency sub-band are obtained based on the transformation of the solar heat collecting plate image, then the texture detail processing result is determined based on the main characteristics of the sub-bands to obtain the foreground pixel point, the texture responsivity and the texture missing degree of the foreground pixel point are further obtained, the defect part is segmented from the foreground image based on the texture missing degree, and finally the production quality result of the heat collecting plate of the solar water heater is determined based on the area ratio of the defect part, so that the accuracy of quality detection of the solar water heater is greatly improved.
The foregoing description is only of the preferred embodiments of the present invention and is not intended to limit the scope of the invention, and all equivalent structures or modifications in the structures or processes described in the specification and drawings, or the direct or indirect application of the present invention to other related technical fields, are included in the scope of the present invention.
Claims (6)
1. A method for detecting production quality of a solar water heater, the method comprising:
performing gray level conversion and wavelet conversion on the acquired solar heat collecting plate image to obtain four sub-bands, wherein each sub-band comprises a low-frequency sub-band and three high-frequency sub-bands;
acquiring a main feature vector of the sub-band, and determining texture detail results of each pixel point based on the main feature vector;
determining foreground pixel points based on texture detail processing results of the pixel points, and calculating texture responsivity of the foreground pixel points;
determining texture missing degree of the foreground pixel point based on the minimum distance value of the foreground pixel point and the edge pixel point and the texture responsivity;
determining a segmentation threshold value based on the texture defect degree of each foreground pixel point, and performing global threshold segmentation on a foreground image corresponding to the foreground pixel point based on the segmentation threshold value to obtain a defect region;
calculating the ratio of the area of the defect area to the area of the foreground image, and determining the heat collecting plate corresponding to the foreground image with the ratio larger than the standard value as a disqualified solar heat collecting plate;
the determining the texture missing degree of the foreground pixel point based on the minimum distance value of the foreground pixel point and the edge pixel point and the texture responsivity comprises the following steps:
calculating the average texture responsivity and the standard deviation of the texture responsivity of each foreground pixel point;
determining the weight of each foreground pixel point based on the average texture responsivity, the texture responsivity standard deviation and the texture missing degree of the foreground pixel point;
determining the product of the texture missing degree, the minimum distance value and the weight of the foreground pixel point as the texture missing degree of the corresponding foreground pixel point;
determining a segmentation threshold based on the texture defect degree of each foreground pixel point, and performing global threshold segmentation on the foreground image corresponding to the foreground pixel point based on the segmentation threshold, wherein the obtaining a defect area comprises:
calculating the edge texture defect degree average value of all the edge pixel points;
taking the Euclidean distance between the foreground pixel point texture missing degree and the edge texture missing degree as a segmentation threshold value, and segmenting the foreground image based on the segmentation threshold value;
and determining the foreground pixel point with the Euclidean distance larger than the experience value as a defective pixel point, and determining the region corresponding to the defective pixel point as a defective region.
2. The method of claim 1, wherein the obtaining the principal eigenvector of the subband and determining texture detail results for each pixel based on the principal eigenvector comprises:
obtaining a vector matrix of the sub-bands, obtaining a first covariance matrix based on the vector matrix, and determining a main feature vector of each sub-band based on a feature value matrix and a feature vector matrix of the first covariance matrix;
and obtaining a main feature vector of the pixel point at the specified position in the coordinate system, and obtaining a texture detail result of the corresponding pixel point based on the main feature vector.
3. The method of claim 2, wherein the obtaining the vector matrix for the subbands, obtaining a first covariance matrix based on the vector matrix, and determining a principal eigenvector for each subband based on an eigenvalue matrix and an eigenvector matrix of the first covariance matrix comprises:
normalizing the pixel value of each pixel point in the sub-band to obtain a normalized pixel value;
constructing a vector matrix of subbands based on the normalized pixel values;
determining vectors in the same direction in the vector matrix as characteristic vectors of subbands, wherein the characteristic vectors are arranged into a characteristic vector matrix according to columns;
compressing eigenvectors of the sub-bands into one based on principal component analysis, and constructing a first covariance matrix of the eigenvector matrix;
performing eigenvalue decomposition on the first covariance matrix to obtain a first eigenvalue and a first eigenvector;
and determining the first eigenvector corresponding to the maximum value in the first eigenvectors as a main eigenvector.
4. The method of claim 2, wherein obtaining the principal eigenvector of the pixel at the specified location in the coordinate system, and obtaining the texture detail result of the corresponding pixel based on the principal eigenvector comprises:
constructing a coordinate system by taking the upper left corner of a subband as an origin, taking the horizontal direction as an x axis and taking the vertical direction as a y axis, and obtaining a main feature vector of a pixel point at a specified position in the coordinate system;
multiplying the principal eigenvector of the pixel point by the transpose of the principal component vector to obtain a high-dimensional vector of the pixel point after inversion;
and normalizing the high-dimensional vector to obtain a texture detail result of the pixel point based on the main feature vector.
5. The method of claim 1, wherein determining foreground pixels based on texture detail processing results for each pixel and calculating texture responsivity for each foreground pixel comprises:
calculating the maximum value of texture detail results of three high-frequency feature vectors of the pixel points, comparing the texture detail results of the low-frequency feature vectors with the maximum value of the texture detail results, and determining the corresponding pixel point as a background pixel point if the texture detail results of the low-frequency feature vectors are larger than the maximum value of the texture detail results;
if the texture detail result of the low-frequency feature vector is smaller than or equal to the maximum value of the texture detail result, determining the corresponding pixel point as a foreground pixel point;
calculating texture responsivity of the foreground pixel points based on texture detail results;
and determining the texture responsivity of the foreground pixel point according to the texture responsivity of the texture detail result.
6. The method of claim 1, wherein prior to determining the texture deficiency of the foreground pixel point based on the minimum distance value of the foreground pixel point from the edge pixel point and the texture responsiveness, further comprising:
performing edge detection on the gray level image through a canny operator to obtain an edge image;
respectively establishing a search window with preset size on the gray level image by taking each pixel point as a center, and gradually increasing the size of the search window until edge pixel points appear in the search window;
and calculating the minimum distance value between each pixel point and the corresponding edge pixel point.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310868108.1A CN116596922B (en) | 2023-07-17 | 2023-07-17 | Production quality detection method of solar water heater |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310868108.1A CN116596922B (en) | 2023-07-17 | 2023-07-17 | Production quality detection method of solar water heater |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116596922A CN116596922A (en) | 2023-08-15 |
| CN116596922B true CN116596922B (en) | 2023-09-12 |
Family
ID=87601200
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310868108.1A Active CN116596922B (en) | 2023-07-17 | 2023-07-17 | Production quality detection method of solar water heater |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116596922B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116740072B (en) * | 2023-08-15 | 2023-12-01 | 安徽省云鹏工程项目管理有限公司 | Road surface defect detection method and system based on machine vision |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012115594A1 (en) * | 2011-02-21 | 2012-08-30 | Stratech Systems Limited | A surveillance system and a method for detecting a foreign object, debris, or damage in an airfield |
| WO2013062944A1 (en) * | 2011-10-26 | 2013-05-02 | The Regents Of The University Of California | Multi view synthesis method and display devices with spatial and inter-view consistency |
| WO2017080420A1 (en) * | 2015-11-09 | 2017-05-18 | Versitech Limited | Auxiliary data for artifacts –aware view synthesis |
| US9760978B1 (en) * | 2016-05-09 | 2017-09-12 | Adobe Systems Incorporated | Missing region prediction |
| CN107301674A (en) * | 2017-05-22 | 2017-10-27 | 东南大学 | A kind of three-dimensional reconstruction reconstruction method based on window interpolation |
| WO2018040118A1 (en) * | 2016-08-29 | 2018-03-08 | 武汉精测电子集团股份有限公司 | Gpu-based tft-lcd mura defect detection method |
| CN109086801A (en) * | 2018-07-06 | 2018-12-25 | 湖北工业大学 | A kind of image classification method based on improvement LBP feature extraction |
| WO2019104767A1 (en) * | 2017-11-28 | 2019-06-06 | 河海大学常州校区 | Fabric defect detection method based on deep convolutional neural network and visual saliency |
| CN112884716A (en) * | 2021-01-28 | 2021-06-01 | 中国空气动力研究与发展中心超高速空气动力研究所 | Method for strengthening characteristics of ultra-high-speed impact damage area |
| CN114049343A (en) * | 2021-11-23 | 2022-02-15 | 沈阳建筑大学 | A retrospective method for complex missing textures in crack propagation process based on deep learning |
| WO2022154787A1 (en) * | 2021-01-13 | 2022-07-21 | Hewlett-Packard Development Company, L.P. | Image region of interest defect detection |
| CN116309157A (en) * | 2023-03-16 | 2023-06-23 | 西北大学 | Qin warrior texture restoration method combining recursive reasoning and feature re-optimization |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5983935B2 (en) * | 2011-11-30 | 2016-09-06 | パナソニックＩｐマネジメント株式会社 | New viewpoint image generation apparatus and new viewpoint image generation method |
| US9881354B2 (en) * | 2012-03-15 | 2018-01-30 | Microsoft Technology Licensing, Llc | Image completion including automatic cropping |
| US9172933B2 (en) * | 2012-10-17 | 2015-10-27 | Samsung Display Co., Ltd. | Correcting anamolous texture and feature width effects in a display that uses a multi primary color unit scheme |
-
2023
- 2023-07-17 CN CN202310868108.1A patent/CN116596922B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012115594A1 (en) * | 2011-02-21 | 2012-08-30 | Stratech Systems Limited | A surveillance system and a method for detecting a foreign object, debris, or damage in an airfield |
| WO2013062944A1 (en) * | 2011-10-26 | 2013-05-02 | The Regents Of The University Of California | Multi view synthesis method and display devices with spatial and inter-view consistency |
| WO2017080420A1 (en) * | 2015-11-09 | 2017-05-18 | Versitech Limited | Auxiliary data for artifacts –aware view synthesis |
| US9760978B1 (en) * | 2016-05-09 | 2017-09-12 | Adobe Systems Incorporated | Missing region prediction |
| WO2018040118A1 (en) * | 2016-08-29 | 2018-03-08 | 武汉精测电子集团股份有限公司 | Gpu-based tft-lcd mura defect detection method |
| CN107301674A (en) * | 2017-05-22 | 2017-10-27 | 东南大学 | A kind of three-dimensional reconstruction reconstruction method based on window interpolation |
| WO2019104767A1 (en) * | 2017-11-28 | 2019-06-06 | 河海大学常州校区 | Fabric defect detection method based on deep convolutional neural network and visual saliency |
| CN109086801A (en) * | 2018-07-06 | 2018-12-25 | 湖北工业大学 | A kind of image classification method based on improvement LBP feature extraction |
| WO2022154787A1 (en) * | 2021-01-13 | 2022-07-21 | Hewlett-Packard Development Company, L.P. | Image region of interest defect detection |
| CN112884716A (en) * | 2021-01-28 | 2021-06-01 | 中国空气动力研究与发展中心超高速空气动力研究所 | Method for strengthening characteristics of ultra-high-speed impact damage area |
| CN114049343A (en) * | 2021-11-23 | 2022-02-15 | 沈阳建筑大学 | A retrospective method for complex missing textures in crack propagation process based on deep learning |
| CN116309157A (en) * | 2023-03-16 | 2023-06-23 | 西北大学 | Qin warrior texture restoration method combining recursive reasoning and feature re-optimization |
Non-Patent Citations (1)
| Title |
|---|
| 纹理合成技术在遥感影像缺失信息恢复中的应用;陈仁喜;李鑫慧;李盛阳;;遥感信息(05);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116596922A (en) | 2023-08-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115861135B (en) | Image enhancement and recognition method applied to panoramic detection of box body | |
| CN107301661B (en) | High-resolution remote sensing image registration method based on edge point features | |
| CN109767439B (en) | Target detection method for multi-scale difference and bilateral filtering of self-adaptive window | |
| CN107945161B (en) | Detection method of road surface defect based on texture feature extraction | |
| CN105956582B (en) | A kind of face identification system based on three-dimensional data | |
| CN116664559B (en) | Machine vision-based memory bank damage rapid detection method | |
| CN105277567B (en) | A kind of fabric defects detection method | |
| CN110766689A (en) | Method and device for detecting article image defects based on convolutional neural network | |
| Krishnan et al. | A survey on different edge detection techniques for image segmentation | |
| CN116137036B (en) | Gene detection data intelligent processing system based on machine learning | |
| CN109636766A (en) | Polarization differential and intensity image Multiscale Fusion method based on marginal information enhancing | |
| CN112308872B (en) | Image Edge Detection Method Based on Multiscale Gabor First Derivative | |
| CN110580705B (en) | Method for detecting building edge points based on double-domain image signal filtering | |
| CN116503393B (en) | Circuit board plasma nano coating quality detection method based on image processing | |
| CN116596922B (en) | Production quality detection method of solar water heater | |
| CN102903108A (en) | Edge detection method based on underwater image statistical property | |
| Liang et al. | Automatic defect detection of texture surface with an efficient texture removal network | |
| CN110348459A (en) | Based on multiple dimensioned quick covering blanket method sonar image fractal characteristic extracting method | |
| Wu et al. | Research on crack detection algorithm of asphalt pavement | |
| CN103942526B (en) | Linear feature extraction method for discrete data point set | |
| CN112329677A (en) | Remote sensing image river target detection method and device based on feature fusion | |
| CN117314901B (en) | A scale-adaptive neural network system for chip detection | |
| CN112927169B (en) | A Noise Removal Method for Remote Sensing Image Based on Wavelet Transform and Improved Weighted Kernel Norm Minimization | |
| CN109598205A (en) | The method of Finger print characteristic abstract and compressed encoding based on Gabor transformation | |
| CN105321178A (en) | Image segmentation method and apparatus based on sparse principal component analysis |
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 |