[go: up one dir, main page]

CN116310360A - Reactor surface defect detection method - Google Patents

Reactor surface defect detection method Download PDF

Info

Publication number
CN116310360A
CN116310360A CN202310558152.2A CN202310558152A CN116310360A CN 116310360 A CN116310360 A CN 116310360A CN 202310558152 A CN202310558152 A CN 202310558152A CN 116310360 A CN116310360 A CN 116310360A
Authority
CN
China
Prior art keywords
cluster
reactor
divided
pixel
gray
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
Application number
CN202310558152.2A
Other languages
Chinese (zh)
Other versions
CN116310360B (en
Inventor
何凯强
何祉辰
何佳泓
何俞瑾
郑微丹
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shide Electric Group Co ltd
Original Assignee
Shide Electric Group Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shide Electric Group Co ltd filed Critical Shide Electric Group Co ltd
Priority to CN202310558152.2A priority Critical patent/CN116310360B/en
Publication of CN116310360A publication Critical patent/CN116310360A/en
Application granted granted Critical
Publication of CN116310360B publication Critical patent/CN116310360B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/30—Noise filtering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/28—Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/36—Applying a local operator, i.e. means to operate on image points situated in the vicinity of a given point; Non-linear local filtering operations, e.g. median filtering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/54—Extraction of image or video features relating to texture
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
    • G06V10/751—Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/761—Proximity, similarity or dissimilarity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
    • G06V10/763—Non-hierarchical techniques, e.g. based on statistics of modelling distributions
    • 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
    • Y02E30/00—Energy generation of nuclear origin
    • Y02E30/30—Nuclear fission reactors

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Computing Systems (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Nonlinear Science (AREA)
  • Image Analysis (AREA)

Abstract

The invention relates to the technical field of image data processing, in particular to a reactor surface defect detection method, which comprises the following steps: dividing the gray level map of the surface of the reactor to obtain a plurality of divided areas, calculating the noise influence degree of each divided area, calculating the density reachable distance, clustering according to the density reachable distance to obtain a plurality of clustering clusters of each divided area, calculating the correction coefficient of each clustering cluster, obtaining the correction noise influence degree of each clustering cluster according to the correction coefficient and the noise influence degree, obtaining the main component directions of all the clustering clusters in all the divided areas, performing filtering operation according to the main component directions of all the clustering clusters in all the divided areas and the correction noise influence degree, obtaining the gray level map of the surface of the reactor after denoising, performing defect recognition, and obtaining the crack defect of the threaded area. The invention reserves texture information in the image, has good denoising effect, can clearly identify defects, and further ensures the delivery quality of the reactor.

Description

Reactor surface defect detection method
Technical Field
The invention relates to the technical field of image data processing, in particular to a method for detecting surface defects of a reactor.
Background
The reactor is also called an inductor, and has wide application in a circuit, and has a certain inductive property because of the electromagnetic induction effect in the circuit, so that the reactor can play a role in preventing current change. When one conductor is electrified, a magnetic field is generated in a certain space occupied by the conductor, and the reactor is formed by winding a wire into a solenoid, so that a core is inserted into the solenoid for enabling the solenoid to have larger inductance, and the core reactor is called as a core reactor. However, in the production process, the surface may be subjected to the action of hard force, so that crack defects are generated, and the quality of products is affected, so that the defect detection is required to be performed on the reactor when the reactor leaves a factory, and the quality of the products is ensured. However, when defects on the surface of a product are identified, noise is generated due to the complex environment in which images are acquired, and the identification of the defects is affected, so that the images need to be subjected to denoising.
In the prior art, a plurality of methods for denoising the image are provided, wherein the bilateral filtering algorithm can effectively remove the influence of noise, and the edges of the image are reserved, so that the textures of the image can be well reserved after denoising. However, when denoising is performed, the fixed filter weight affects the denoising effect of the algorithm, so that a proper filter weight needs to be obtained according to detail change in the image. The denoising effect of the artificially set filtering weights is not good, and adaptive filtering weights are required.
Disclosure of Invention
The invention provides a method for detecting surface defects of a reactor, which aims to solve the existing problems.
The invention discloses a method for detecting surface defects of a reactor, which adopts the following technical scheme:
one embodiment of the invention provides a method for detecting surface defects of a reactor, which comprises the following steps:
collecting a reactor surface image; dividing a gray scale image of the surface of the reactor to obtain a plurality of divided areas;
obtaining the noise influence degree of each divided area according to the difference of gray values of the pixel points in the minimum circumscribed rectangle and the neighborhood pixel points of each divided area and the distribution direction of the minimum circumscribed rectangle;
dividing the gray scale range of the gray scale map of the surface of the reactor into three sections on average, and calculating the density reachable distance of the gray scale map of the surface of the reactor according to the difference of noise influence degrees of all the divided sections and the distribution condition of all the pixel points corresponding to each section; clustering each segmented region according to the density reachable distance to obtain a plurality of clustering clusters of each segmented region;
obtaining a correction coefficient of each cluster according to the number of the pixel points of the cluster after extension and the gray value difference between all the pixel points and the neighborhood pixel points;
recording the product of the correction coefficient of each cluster in each partition area and the noise influence degree of each partition area as the correction noise influence degree of each cluster in each partition area;
analyzing the vertical direction of the maximum difference value of all pixel points of each cluster in each partition area to obtain the main component direction of each cluster in each partition area;
and performing filtering operation according to the main component directions of all clusters in all the partitioned areas and the influence degree of the corrected noise to obtain a denoised reactor surface gray scale image.
Further, the obtaining the noise influence degree of each divided area includes the following specific steps:
obtaining the minimum circumscribed rectangle of each divided area, marking the longer side of the minimum circumscribed rectangle as a long side, marking the shorter side as a short side, and constructing a rectangular coordinate system by taking the long side of each minimum circumscribed rectangle as a transverse axis and the short side as a longitudinal axis; the calculation formula of the noise influence degree of any one of the divided regions is as follows:
Figure SMS_1
in the method, in the process of the invention,
Figure SMS_4
represent the first
Figure SMS_7
The degree of noise influence of the individual divided areas,
Figure SMS_10
、
Figure SMS_2
respectively represent the first
Figure SMS_5
The length of the long side and short side of the smallest bounding rectangle of each divided area,
Figure SMS_8
represent the first
Figure SMS_11
The gray value of the pixel point with x abscissa and y ordinate in the minimum circumscribed rectangle of each divided area,
Figure SMS_3
represent the first
Figure SMS_6
The slope of the long side of the smallest bounding rectangle of the individual segmented regions,
Figure SMS_9
represent the first
Figure SMS_12
Gray values of the ith pixel point in eight adjacent areas of the pixel points with x horizontal coordinates and y vertical coordinates in the minimum circumscribed rectangle of each divided area;
obtaining the noise influence degree of all the divided areas, carrying out linear normalization on the noise influence degree of all the divided areas, and taking the noise influence degree after the linear normalization as the noise influence degree of each divided area.
Further, the calculating the density reachable distance of the gray map of the surface of the reactor comprises the following specific steps:
the calculation formula of the density reachable distance of the gray map of the surface of the reactor is as follows:
Figure SMS_13
in the method, in the process of the invention,
Figure SMS_15
the density of the gray scale representing the reactor surface can be a distance,
Figure SMS_18
represents the number of possible cases where two divided regions are selected from among all divided regions, t represents the number of divided regions,
Figure SMS_20
、
Figure SMS_16
the noise influence degrees of the divided areas a and b are respectively expressed,
Figure SMS_17
the representation takes the absolute value of the value,
Figure SMS_19
representing the number of possible cases where two pixel points are selected from all pixel points corresponding to the j-th interval,
Figure SMS_21
indicating the number of pixels corresponding to the j-th interval,
Figure SMS_14
and the Euclidean distance between the pixel point m and the pixel point n in all the pixel points corresponding to the j-th interval is represented.
Further, the obtaining a plurality of clusters of each segmented region includes the following specific steps:
and clustering each segmented region by using the density reachable distance of the reactor surface gray map as a density reachable distance parameter of DBSCAN density clustering, and obtaining a plurality of clustering clusters of each segmented region by using a DBSCAN density clustering algorithm.
Further, the obtaining the correction coefficient of each cluster includes the following specific steps:
and (3) extending each obtained cluster outwards by 5 pixel areas, taking the extended areas as areas corresponding to each cluster, wherein the calculation formula of the correction coefficient of any cluster is as follows:
Figure SMS_22
in the method, in the process of the invention,
Figure SMS_38
represent the first
Figure SMS_42
The first of the divided regions
Figure SMS_45
The correction coefficients of the individual clusters are used,
Figure SMS_25
represent the first
Figure SMS_27
The first of the divided regions
Figure SMS_31
The number of pixels of the cluster,
Figure SMS_34
represent the first
Figure SMS_39
The number of pixels of the individual divided areas,
Figure SMS_43
represent the first
Figure SMS_47
The first of the divided regions
Figure SMS_49
The number of pixels in a cluster,
Figure SMS_41
represent the first
Figure SMS_46
The first of the divided regions
Figure SMS_48
The gray value of the q-th pixel point of the cluster,
Figure SMS_50
represent the first
Figure SMS_29
The first of the divided regions
Figure SMS_36
The gray value of the pixel point on the left side of the q-th pixel point of the cluster,
Figure SMS_40
represent the first
Figure SMS_44
The first of the divided regions
Figure SMS_23
The gray value of the pixel to the right of the q-th pixel of the cluster,
Figure SMS_30
represent the first
Figure SMS_33
The first of the divided regions
Figure SMS_37
The gray value of the pixel above the q-th pixel of the cluster,
Figure SMS_24
represent the first
Figure SMS_28
The first of the divided regions
Figure SMS_32
The gray value of the pixel below the q-th pixel of the cluster,
Figure SMS_35
an exponential function based on natural constants is represented,
Figure SMS_26
the representation takes absolute value.
Further, the obtaining the principal component direction of each cluster in each partition area includes the following specific steps:
marking any cluster in any one partitioned area as a target cluster, and obtaining the gradient direction and the gradient amplitude of each pixel point in the target cluster through a Sobel operator; the vector taking the gradient direction of the pixel point as the direction and taking the gradient amplitude of the pixel point as the modulus is marked as the direction vector of the pixel point;
taking any pixel point in the target cluster as a target pixel point, acquiring an absolute value of a difference value between the gradient direction of the target pixel point and the gradient direction of each neighborhood pixel point in the neighborhood of the target pixel point 4, and marking the absolute value as the gradient direction variation quantity of the target pixel point; adding the direction vector of the neighborhood pixel point corresponding to the maximum gradient direction variation and the direction vector of the target pixel point, and marking the vertical direction of the obtained vector as the vertical direction of the maximum difference value of the target pixel point;
and obtaining the vertical direction of the maximum difference value of all pixel points in the target cluster, and carrying out principal component analysis on the vertical direction of all maximum difference values to obtain the principal component direction of the target cluster.
Further, the method for obtaining the denoised reactor surface gray scale map comprises the following specific steps:
and taking the main component direction of each cluster in each partition area as a filtering direction, taking the modified noise influence degree of each cluster in each partition area as a filtering weight, and carrying out filtering operation on the area corresponding to each cluster in each partition area according to a bilateral filtering algorithm to obtain a denoised reactor surface gray scale map.
The technical scheme of the invention has the beneficial effects that: when the defect detection is carried out on the surface of the reactor, the complete defect area cannot be accurately segmented because the image is affected by noise, so that the image is subjected to self-adaptive denoising through a bilateral filtering algorithm. The de-noised image can retain texture information in the image, has a good de-noising effect, and can clearly identify defects according to the de-noised image, so that the factory quality of the reactor is ensured.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart showing steps of a method for detecting surface defects of a reactor according to the present invention;
FIG. 2 is a partial surface image of a reactor including crack defects according to one embodiment of the present invention;
FIG. 3 is a view of the denoised image of FIG. 2 according to one embodiment of the present invention;
fig. 4 is a schematic diagram of a crack defect detection according to an embodiment of the present invention.
Detailed Description
In order to further describe the technical means and effects adopted by the invention to achieve the preset aim, the following detailed description is given below of a method for detecting surface defects of a reactor according to the invention, which is specific to the implementation, structure, characteristics and effects thereof, with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The following specifically describes a specific scheme of the method for detecting the surface defects of the reactor provided by the invention with reference to the accompanying drawings.
Referring to fig. 1, a flowchart of steps of a method for detecting a surface defect of a reactor according to an embodiment of the invention is shown, the method includes the following steps:
s001, acquiring an image of the surface of the reactor.
It should be noted that, the main purpose of the present invention is to identify defects on the surface of the reactor, so that it is necessary to first acquire an image of the surface of the reactor.
In this embodiment, because the detected area is a coil area of the reactor, the reactor needs to be horizontally placed during acquisition, a CCD industrial camera is used to shoot the surface of the reactor in a top view, and then the acquired image is subjected to grayscale processing to obtain a surface image of the reactor; the image is grayed by using an average method, which is a known technique and will not be described in detail herein. Referring to fig. 2, a partial surface image of a reactor including a crack defect according to an embodiment of the present invention is shown. As can be seen from fig. 2, the overall graininess of the image is severe due to the influence of noise, and the noise with a low gray value affects the recognition of crack defects.
Thus, a reactor surface gray scale map is obtained.
S002, segmenting the gray level map of the surface of the reactor to obtain a plurality of segmented areas, calculating the noise influence degree of each segmented area, calculating the density reachable distance of the gray level map of the surface of the reactor, clustering each segmented area according to the density reachable distance to obtain a plurality of clusters of each segmented area, calculating the correction coefficient of each cluster in each segmented area, and obtaining the correction noise influence degree of each cluster according to the correction coefficient and the noise influence degree.
It should be noted that, due to the influence of the collecting environment, there is much noise in the collected surface image of the reactor, which affects the accuracy of the defect detection result, so that the denoising process is required for the surface image of the reactor. The bilateral filtering algorithm comprehensively considers the gray value and the space distance of the pixel point during denoising, combines the gray value and the space distance and carries out compromise processing, thus not only effectively reducing the influence of various noises in the image, but also protecting the edge information in the image. The bilateral filtering denoising algorithm is a denoising method through Gaussian kernel convolution, but when denoising is performed, because noise influence degrees of different areas are different, and because local gray level differences are increased due to original textures, adjacent pixels are screened according to the similarity of gray values in adjacent pixel set templates, pixels with similar gray values are selected to form a new pixel set template, and then optimal filtering weights are obtained according to pixel changes of images.
It should be further noted that, the noise exists in the image, which is equivalent to adding a part of pixel points to the original image, and the model is as follows: noise image = original image + noise image, and noise will change the local gray scale of original image, and in the present invention, the noise that the image exists is mostly gaussian noise, also will exist with a small part of pretzel noise, gaussian noise is the pixel that the gray scale value is great, and the gray scale value of pretzel noise pixel is possible to be great, and possible to be less. And when an image is acquired, the plastic on the coil can generate local reflection to influence the recognition of noise, so that a pixel projection model is constructed according to the distribution of pixel points in the image to analyze the distribution of noise.
1. And dividing the gray level diagram of the surface of the reactor to obtain a plurality of dividing areas.
The invention relates to a method for dividing an image into areas, which is characterized in that gray values of different pixels are displayed on a plane, when noise exists in the image, the noise pixels are randomly distributed, the continuity characteristic of the gray values of the pixels in each area of the image is changed, and because coils wound on the surface of a reactor have texture characteristics, the pixels in different space positions have certain regularity, when the image is divided into areas, the gray values are light and dark gray bands with gray alternating changes, the middle gray value of each gray band is large, the gray values of the two sides are small, and then the noise influence degree of each local area is calculated.
In the embodiment, a watershed algorithm is adopted to segment a gray level diagram of the surface of the reactor, and a plurality of segmentation areas are obtained. The watershed algorithm is a well-known technology and will not be described in detail herein.
In this embodiment, the flooding threshold of the watershed algorithm is 25, which is an empirical threshold, and the implementer can set the threshold according to different real-time environments.
2. The noise influence degree of each divided region is calculated.
It should be noted that, since the surface of the reactor has a crack defect, after the reactor is segmented by the watershed algorithm, the texture of the surface and the crack defect are mixed, and thus, when determining the influence degree of the noise of the image through the neighborhood pixel points of the pixel points, the directions of the crack and the texture need to be considered. Because the details of the image are not lost when denoising is performed, and the denoising effect is good, the change of the crack area needs to be highlighted when the analysis is performed according to the neighborhood pixel points. Since the textures between the reactor coils are regularly distributed and the directions of the variation of the cracks are random, the distribution direction of the determined pixel points is obtained from the minimum bounding rectangle.
In this embodiment, a minimum circumscribed rectangle of each divided area is obtained, a longer side of the minimum circumscribed rectangle is marked as a long side, a shorter side is marked as a short side, and a long side of each minimum circumscribed rectangle is marked as a horizontal axis and a short side is marked as a vertical axis, so as to construct a rectangular coordinate system; according to the difference of gray values of pixel points in the minimum bounding rectangle and the neighborhood pixel points of each divided area and the distribution direction of the minimum bounding rectangle, the noise influence degree of each divided area is obtained, and the calculation formula of the noise influence degree of any divided area is as follows:
Figure SMS_51
in the method, in the process of the invention,
Figure SMS_54
represent the first
Figure SMS_57
The degree of noise influence of the individual divided areas,
Figure SMS_60
、
Figure SMS_53
respectively represent the first
Figure SMS_56
The length of the long side and short side of the smallest bounding rectangle of each divided area,
Figure SMS_59
represent the first
Figure SMS_62
The gray value of the pixel point with x abscissa and y ordinate in the minimum circumscribed rectangle of each divided area,
Figure SMS_52
represent the first
Figure SMS_55
The slope of the long side of the smallest bounding rectangle of the individual segmented regions,
Figure SMS_58
represent the first
Figure SMS_61
And gray values of the ith pixel point in eight adjacent pixel points of which the abscissa is x and the ordinate is y in the minimum circumscribed rectangle of each divided region.
Obtaining the noise influence degree of all the divided areas, carrying out linear normalization on the noise influence degree of all the divided areas, and taking the noise influence degree after the linear normalization as the noise influence degree of each divided area.
Since the grain direction of the crack is random and the grain direction on the image is fixed, the grain direction in the divided areas is represented by the slope of the smallest circumscribed rectangle where the divided areas are located, the gray scale change of the pixel points in each divided area is continuous after the image is divided, when random noise exists, the gray scale value of the adjacent pixel points is suddenly changed, the noise is evaluated according to the gray scale value change of the adjacent pixel points, and since the grain directions of the crack and the image are different, when difference calculation is performed, the gray scale difference value change between the adjacent pixel points is larger, and the gray scale change of other textures is smaller, so that the crack area can be evaluated for emphasis.
3. And calculating the density reachable distance of the reactor surface gray level graph, and clustering each segmented region according to the density reachable distance to obtain a plurality of clustering clusters of each segmented region.
Since the degree of influence of noise in each region is obtained by the calculation, and since the entire evaluation is performed for each divided region, when there is a crack defect in the divided region, the gradation continuity between pixels is changed, and the gradation value between adjacent pixels is changed, the obtained degree of influence of noise is larger than the actual value, and therefore the degree of influence of noise is corrected according to the edge characteristic change of the crack. In the region division, the region of each wire is obtained, and in order to accurately describe the influence of noise in an image, the image is divided. Because the gray value of the crack area is smaller than the gray value of the wire area when the crack occurs, the obtained segmented area is subjected to DBSCAN density clustering, because the density clustering is performed according to the distance between the pixel points, the gray value of the image is divided when the clustering is performed, the gray value of the image is divided into three sections on average, then the density clustering is performed, different clustering clusters are obtained, and when the density clustering is performed, the density reachable distance needs to be set in advance, but in the invention, the noise influence degree of the local area is possibly larger because of the influence of noise, and the density reachable distance is set smaller because of the fact that the segmentation of the area is inaccurate when the clustering is performed, so that the density reachable distance needs to be determined according to the influence degree of the noise.
In the present embodiment, the range formed by the smallest gray value and the largest gray value in the reactor surface gray map is referred to as the gray range of the reactor surface gray map, and the gray range is divided into three sections on average, which are referred to as the 1 st section to the 3 rd section, respectively.
According to the difference of noise influence degrees of the use division areas and the distribution condition of all pixel points corresponding to each interval, calculating the density reachable distance of the gray map of the surface of the reactor, wherein the calculation formula of the density reachable distance of the gray map of the surface of the reactor is as follows:
Figure SMS_63
in the method, in the process of the invention,
Figure SMS_66
the density of the gray scale representing the reactor surface can be a distance,
Figure SMS_68
represents the number of possible cases where two divided regions are selected from among all divided regions, t represents the number of divided regions,
Figure SMS_70
、
Figure SMS_65
the noise influence degrees of the divided areas a and b are respectively expressed,
Figure SMS_67
the representation takes the absolute value of the value,
Figure SMS_69
representing the number of possible cases where two pixel points are selected from all pixel points corresponding to the j-th interval,
Figure SMS_71
indicating the number of pixels corresponding to the j-th interval,
Figure SMS_64
and the Euclidean distance between the pixel point m and the pixel point n in all the pixel points corresponding to the j-th interval is represented.
Figure SMS_72
The mean value of the difference of the noise influence degrees of every two divided areas is represented, and the noise influence degrees of different divided areas are different when density clustering is carried out, so that the density reachable distance is adjusted by adopting the average noise influence degree in order to realize clustering on all the divided areas in a balanced mode. Since the density reachable distance is determined according to the distance between the pixel points, the density reachable distance is determined by calculating the distance average value of all the pixel points in each section.
Setting the minimum cluster number parameter of DBSCAN density clustering as 4, taking the density reachable distance of the reactor surface gray map as the density reachable distance parameter of DBSCAN density clustering, and clustering each segmented region through a DBSCAN density clustering algorithm to obtain a plurality of cluster clusters of each segmented region. The DBSCAN density clustering algorithm is a prior known technology and is not described in detail herein.
4. And calculating a correction coefficient of each cluster in each partition area, and obtaining the correction noise influence degree of each cluster according to the correction coefficient and the noise influence degree.
In this embodiment, in order to more accurately reflect the association of clusters in the partitioned area, each obtained cluster is extended outwards by 5 pixel areas, the extended area is taken as the area corresponding to each cluster, and the correction coefficient of each cluster is obtained according to the number of pixels of the extended cluster and the gray value difference between all pixels and the neighbor pixels; the calculation formula of the correction coefficient of any cluster is as follows:
Figure SMS_73
in the method, in the process of the invention,
Figure SMS_92
represent the first
Figure SMS_96
The first of the divided regions
Figure SMS_99
The correction coefficients of the individual clusters are used,
Figure SMS_75
represent the first
Figure SMS_80
The first of the divided regions
Figure SMS_83
The number of pixels of the cluster,
Figure SMS_87
represent the first
Figure SMS_90
The number of pixels of the individual divided areas,
Figure SMS_94
represent the first
Figure SMS_97
The first of the divided regions
Figure SMS_100
The number of pixels in a cluster,
Figure SMS_91
represent the first
Figure SMS_95
The first of the divided regions
Figure SMS_98
The gray value of the q-th pixel point of the cluster,
Figure SMS_101
represent the first
Figure SMS_76
The first of the divided regions
Figure SMS_78
The gray value of the pixel point on the left side of the q-th pixel point of the cluster,
Figure SMS_82
represent the first
Figure SMS_86
The first of the divided regions
Figure SMS_74
The gray value of the pixel to the right of the q-th pixel of the cluster,
Figure SMS_79
represent the first
Figure SMS_85
The first of the divided regions
Figure SMS_88
The gray value of the pixel above the q-th pixel of the cluster,
Figure SMS_77
represent the first
Figure SMS_81
The first of the divided regions
Figure SMS_84
The gray value of the pixel below the q-th pixel of the cluster,
Figure SMS_89
an exponential function based on natural constants is represented,
Figure SMS_93
the representation takes absolute value.
Figure SMS_102
The change condition of gray values of the pixel points and the neighborhood pixel points is reflected, if cracks exist in the cluster, the gray value change along the crack direction is similar, but the gray difference of the pixel points along the vertical direction of the crack direction is larger; when the pixel points are noise points, the gray value changes of the neighborhood pixel points in the four surrounding neighbors are similar, and whether the cluster has the influence of cracks or not is reflected according to the ratio of the gray value changes of the neighborhood pixel points. When the ratio is close to 1, it is indicated that the pixel is affected by noise, and when the ratio difference is larger, it is indicated that the degree of influence of noise due to the influence of cracks is larger.
Figure SMS_103
Representing the relative size of the clustered regions because the number of pixels is greater in the clustered regions, but the noise pixels are relatively independentEven if some noise pixels are grouped into one type, the number of the pixels is smaller than that of the crack region, so that the probability of the crack region is expressed by the ratio of the number of the pixels.
The influence degree of noise is corrected according to the change characteristics of clusters in a segmented region in an image, the distribution of noise in a local region can be accurately reflected, the increase of noise evaluation indexes caused by textures existing in an original image can be avoided when denoising is performed, the texture details existing in the image can be reserved when denoising is performed through bilateral filtering, the influence of noise can be well removed, and further the recognition of the crack defects on the surface of a reactor is more accurate.
In the correction, the greater the possibility of cracks, the more the gray level of the pixel changes, and the smaller the actual noise influence, so that the correction parameter obtained directly from the gray level change of the different areas is multiplied by the original noise influence, i.e., the noise influence of the current area can be represented.
The correction parameters of the noise influence degrees of different areas are calculated through the method, then the noise influence degrees of different areas are corrected, and the product of the correction coefficient of each cluster in each partition area and the noise influence degree of each partition area is recorded as the correction noise influence degree of each cluster in each partition area.
S003, obtaining the main component directions of all clusters in all the divided areas.
It should be noted that, the noise influence degrees of different areas in the image are obtained through calculation by the method, and then the sliding direction of the filter window is determined according to the gray level change direction in the image. Since the blurring of details is different from the filtering direction, when filtering is performed along the crack direction, the edge blurring degree of the crack is small, and thus the denoising filtering direction is determined according to the clustering result.
In the embodiment, any cluster in any one partition area is marked as a target cluster, and the gradient direction and the gradient amplitude of each pixel point in the target cluster are obtained through a Sobel operator; a vector in which the gradient direction of the pixel is taken as a direction and the gradient amplitude of the pixel is taken as a modulo vector is taken as a direction vector of the pixel.
Taking any pixel point in the target cluster as a target pixel point, acquiring an absolute value of a difference value between the gradient direction of the target pixel point and the gradient direction of each neighborhood pixel point in the neighborhood of the target pixel point 4, and marking the absolute value as the gradient direction variation quantity of the target pixel point; and carrying out addition operation on the direction vector of the neighborhood pixel point corresponding to the maximum gradient direction variation and the direction vector of the target pixel point, and marking the vertical direction of the obtained vector as the vertical direction of the maximum difference value of the target pixel point.
Obtaining the maximum difference vertical direction of all pixel points in the target cluster, and carrying out principal component analysis on all the maximum difference vertical directions to obtain the principal component direction of the target cluster; and obtaining the principal component directions of all clusters in all the partitioned areas. It should be noted that, according to the principal component analysis algorithm, a plurality of principal component directions are obtained together, and each principal component direction vector corresponds to a feature value, and in this embodiment, only the principal component direction with the largest feature value is retained.
S004, performing filtering operation according to the main component directions of all clusters in all the partition areas and the influence degree of correction noise, obtaining a denoised reactor surface gray scale map, performing defect identification, and obtaining crack defects of the thread areas.
Taking the main component direction of each cluster in each partition area as a filtering direction, taking the modified noise influence degree of each cluster in each partition area as a filtering weight, and carrying out filtering operation on the area corresponding to each cluster in each partition area according to a bilateral filtering algorithm to obtain a denoised reactor surface gray scale map; the bilateral filtering algorithm is a known technology, and will not be described in detail herein. Referring to fig. 3, a diagram of the denoised image of fig. 2 according to an embodiment of the present invention is shown. As can be seen from fig. 3, after denoising, the whole image is smoother, and the influence of noise in fig. 2 is avoided.
And (3) carrying out threshold segmentation on the denoised reactor surface gray level diagram, carrying out defect identification, obtaining crack defects of a threaded region, and setting the threshold range as [20,50]. Referring to fig. 4, which is a schematic diagram of detecting a crack defect according to an embodiment of the present invention, after threshold segmentation in fig. 3, a binary image shown in fig. 4 may be obtained, and corresponding pixels in the binary image may be the crack defect pixels and may be mapped directly into an original image for defect positioning.
When the defect detection is carried out on the surface of the reactor, the complete defect area cannot be accurately segmented because the image is affected by noise, so that the image is subjected to self-adaptive denoising through a bilateral filtering algorithm. The de-noised image can retain texture information in the image, has a good de-noising effect, and can clearly identify defects according to the de-noised image, so that the factory quality of the reactor is ensured.
It should be noted that: the sequence of the embodiments of the present invention is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments.

Claims (7)

1. A method for detecting surface defects of a reactor, comprising the steps of:
collecting a reactor surface image; dividing a gray scale image of the surface of the reactor to obtain a plurality of divided areas;
obtaining the noise influence degree of each divided area according to the difference of gray values of the pixel points in the minimum circumscribed rectangle and the neighborhood pixel points of each divided area and the distribution direction of the minimum circumscribed rectangle;
dividing the gray scale range of the gray scale map of the surface of the reactor into three sections on average, and calculating the density reachable distance of the gray scale map of the surface of the reactor according to the difference of noise influence degrees of all the divided sections and the distribution condition of all the pixel points corresponding to each section; clustering each segmented region according to the density reachable distance to obtain a plurality of clustering clusters of each segmented region;
obtaining a correction coefficient of each cluster according to the number of the pixel points of the cluster after extension and the gray value difference between all the pixel points and the neighborhood pixel points;
recording the product of the correction coefficient of each cluster in each partition area and the noise influence degree of each partition area as the correction noise influence degree of each cluster in each partition area;
analyzing the vertical direction of the maximum difference value of all pixel points of each cluster in each partition area to obtain the main component direction of each cluster in each partition area;
and performing filtering operation according to the main component directions of all clusters in all the partition areas and the influence degree of correction noise to obtain a denoised reactor surface gray scale image, and performing defect identification to obtain the crack defect of the thread area.
2. The method for detecting surface defects of a reactor according to claim 1, wherein the obtaining of the noise influence degree of each divided region comprises the specific steps of:
obtaining the minimum circumscribed rectangle of each divided area, marking the longer side of the minimum circumscribed rectangle as a long side, marking the shorter side as a short side, and constructing a rectangular coordinate system by taking the long side of each minimum circumscribed rectangle as a transverse axis and the short side as a longitudinal axis; the calculation formula of the noise influence degree of any one of the divided regions is as follows:
Figure QLYQS_1
in the method, in the process of the invention,
Figure QLYQS_3
indicate->
Figure QLYQS_6
Noise influence degree of individual divided regions, +.>
Figure QLYQS_9
、/>
Figure QLYQS_4
Respectively represent +.>
Figure QLYQS_7
Length of long side and short side of minimum bounding rectangle of each divided region, < >>
Figure QLYQS_10
Indicate->
Figure QLYQS_12
Gray values of pixel points with x abscissa and y ordinate in minimum circumscribed rectangle of each divided area, +.>
Figure QLYQS_2
Indicate->
Figure QLYQS_5
The slope of the long side of the smallest bounding rectangle of the individual segmented regions,
Figure QLYQS_8
indicate->
Figure QLYQS_11
Gray values of the ith pixel point in eight adjacent areas of the pixel points with x horizontal coordinates and y vertical coordinates in the minimum circumscribed rectangle of each divided area;
obtaining the noise influence degree of all the divided areas, carrying out linear normalization on the noise influence degree of all the divided areas, and taking the noise influence degree after the linear normalization as the noise influence degree of each divided area.
3. The method for detecting surface defects of a reactor according to claim 1, wherein the calculating the density reachable distance of the gray map of the surface of the reactor comprises the following specific steps:
the calculation formula of the density reachable distance of the gray map of the surface of the reactor is as follows:
Figure QLYQS_13
in the method, in the process of the invention,
Figure QLYQS_16
density reachable distance of gray-scale map representing reactor surface,/->
Figure QLYQS_17
Represents the number of possible cases of selecting two divided regions from all divided regions, t represents the number of divided regions, +.>
Figure QLYQS_19
、/>
Figure QLYQS_15
Representing the degree of influence of noise of the divided areas a and b, respectively, +.>
Figure QLYQS_18
Representing absolute value>
Figure QLYQS_20
Representing the number of possible cases of selecting two pixels from all pixels corresponding to the jth interval,/->
Figure QLYQS_21
Indicates the number of pixel points corresponding to the j-th interval, < >>
Figure QLYQS_14
And the Euclidean distance between the pixel point m and the pixel point n in all the pixel points corresponding to the j-th interval is represented.
4. The method for detecting surface defects of a reactor according to claim 1, wherein the obtaining a plurality of clusters of each divided region comprises the specific steps of:
and clustering each segmented region by using the density reachable distance of the reactor surface gray map as a density reachable distance parameter of DBSCAN density clustering, and obtaining a plurality of clustering clusters of each segmented region by using a DBSCAN density clustering algorithm.
5. The method for detecting surface defects of a reactor according to claim 1, wherein the step of obtaining the correction coefficient of each cluster comprises the steps of:
and (3) extending each obtained cluster outwards by 5 pixel areas, taking the extended areas as areas corresponding to each cluster, wherein the calculation formula of the correction coefficient of any cluster is as follows:
Figure QLYQS_22
in the method, in the process of the invention,
Figure QLYQS_39
indicate->
Figure QLYQS_43
The>
Figure QLYQS_46
Correction factors for the individual clusters, +.>
Figure QLYQS_24
Indicate->
Figure QLYQS_27
The>
Figure QLYQS_31
The number of pixels of the cluster, +.>
Figure QLYQS_35
Indicate->
Figure QLYQS_41
The number of pixels of the individual divided areas, < >>
Figure QLYQS_44
Indicate->
Figure QLYQS_47
The>
Figure QLYQS_49
The number of pixels in each cluster, +.>
Figure QLYQS_42
Indicate->
Figure QLYQS_45
The>
Figure QLYQS_48
Gray value of the q-th pixel of the cluster,>
Figure QLYQS_50
indicate->
Figure QLYQS_25
The>
Figure QLYQS_29
Gray value of pixel point on left side of q-th pixel point of each cluster,/>
Figure QLYQS_33
Indicate->
Figure QLYQS_37
The>
Figure QLYQS_23
Gray value of pixel on right side of the q-th pixel of the cluster, +.>
Figure QLYQS_28
Indicate->
Figure QLYQS_32
The>
Figure QLYQS_36
Gray value of pixel above the q-th pixel of the cluster,/->
Figure QLYQS_26
Indicate->
Figure QLYQS_30
The>
Figure QLYQS_34
Gray value of pixel below the q-th pixel of the cluster,/>
Figure QLYQS_38
Representing a base number based on natural constantsExponential function of>
Figure QLYQS_40
The representation takes absolute value.
6. The method for detecting surface defects of a reactor according to claim 1, wherein the step of obtaining the principal component direction of each cluster in each divided region comprises the steps of:
marking any cluster in any one partitioned area as a target cluster, and obtaining the gradient direction and the gradient amplitude of each pixel point in the target cluster through a Sobel operator; the vector taking the gradient direction of the pixel point as the direction and taking the gradient amplitude of the pixel point as the modulus is marked as the direction vector of the pixel point;
taking any pixel point in the target cluster as a target pixel point, acquiring an absolute value of a difference value between the gradient direction of the target pixel point and the gradient direction of each neighborhood pixel point in the neighborhood of the target pixel point 4, and marking the absolute value as the gradient direction variation quantity of the target pixel point; adding the direction vector of the neighborhood pixel point corresponding to the maximum gradient direction variation and the direction vector of the target pixel point, and marking the vertical direction of the obtained vector as the vertical direction of the maximum difference value of the target pixel point;
and obtaining the vertical direction of the maximum difference value of all pixel points in the target cluster, and carrying out principal component analysis on the vertical direction of all maximum difference values to obtain the principal component direction of the target cluster.
7. The method for detecting surface defects of a reactor according to claim 1, wherein the step of obtaining the denoised surface gray scale map of the reactor comprises the following specific steps:
and taking the main component direction of each cluster in each partition area as a filtering direction, taking the modified noise influence degree of each cluster in each partition area as a filtering weight, and carrying out filtering operation on the area corresponding to each cluster in each partition area according to a bilateral filtering algorithm to obtain a denoised reactor surface gray scale map.
CN202310558152.2A 2023-05-18 2023-05-18 Reactor surface defect detection method Active CN116310360B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310558152.2A CN116310360B (en) 2023-05-18 2023-05-18 Reactor surface defect detection method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310558152.2A CN116310360B (en) 2023-05-18 2023-05-18 Reactor surface defect detection method

Publications (2)

Publication Number Publication Date
CN116310360A true CN116310360A (en) 2023-06-23
CN116310360B CN116310360B (en) 2023-08-18

Family

ID=86803539

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310558152.2A Active CN116310360B (en) 2023-05-18 2023-05-18 Reactor surface defect detection method

Country Status (1)

Country Link
CN (1) CN116310360B (en)

Cited By (24)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116994352A (en) * 2023-08-02 2023-11-03 河北科华防静电地板制造有限公司 Computer lab regularly patrol and examine monitoring system based on intelligence antistatic floor
CN117058670A (en) * 2023-10-12 2023-11-14 深圳市华加生物科技有限公司 Electronic tobacco tar flexibility evaluation method
CN117078667A (en) * 2023-10-13 2023-11-17 山东克莱蒙特新材料科技有限公司 Mineral casting detection method based on machine vision
CN117173703A (en) * 2023-11-02 2023-12-05 温州华嘉电器有限公司 Isolating switch state identification method
CN117292328A (en) * 2023-11-24 2023-12-26 山东新中鲁建设有限公司 Safety management and monitoring method and system for construction quality of assembled building
CN117314899A (en) * 2023-11-28 2023-12-29 深圳市烯碳复合材料有限公司 Carbon fiber plate quality detection method based on image characteristics
CN117392469A (en) * 2023-12-11 2024-01-12 深圳市丰源升科技有限公司 Perovskite battery surface coating detection method and system based on machine vision
CN117426776A (en) * 2023-12-20 2024-01-23 河北雄安九康科技有限公司 Electrocardiogram signal characteristic intelligent extraction method
CN117455915A (en) * 2023-12-25 2024-01-26 陕西福坤顺科技有限公司 Machine vision-based three-axis turntable defect detection method
CN117115494B (en) * 2023-10-23 2024-02-06 卡松科技股份有限公司 Lubricating oil impurity pollution detection method and device based on artificial intelligence
CN117764967A (en) * 2023-12-28 2024-03-26 山西建投集团装饰有限公司 Fitment quality detection method and system based on artificial intelligence
CN117853493A (en) * 2024-03-08 2024-04-09 山东天意机械股份有限公司 Wallboard production quality detection method based on image processing
CN117876359A (en) * 2024-03-04 2024-04-12 金品计算机科技(天津)有限公司 Rubber product defect detection method, device, equipment and medium
CN117876402A (en) * 2024-03-13 2024-04-12 中国人民解放军总医院第一医学中心 An intelligent segmentation method for temporomandibular joint disorder images
CN117934469A (en) * 2024-03-22 2024-04-26 张家港市升恒机械设备制造有限公司 Visual detection method for production quality of mining high-voltage frequency converter shell
CN117952860A (en) * 2024-03-27 2024-04-30 山东正禾大教育科技有限公司 Mobile digital publishing method and system
CN118014882A (en) * 2024-04-08 2024-05-10 汉中精测电器有限责任公司 Drilling positioning method of torque sensor based on image processing
CN118196082A (en) * 2024-05-13 2024-06-14 中铁七局集团第三工程有限公司 Method and system for plugging civil air defense channel
CN118483195A (en) * 2024-05-17 2024-08-13 陕西天域新型环保建材有限公司 Detection method of water retention performance of special gypsum mortar based on optical analysis
CN118823024A (en) * 2024-09-20 2024-10-22 东莞市隆慧电子科技有限公司 A method for identifying machining defects of air-cooled radiators
CN118864470A (en) * 2024-09-26 2024-10-29 瓦拉赫(大连)智能科技有限公司 An intelligent analysis system for obstetric ultrasound examination data
CN119006503A (en) * 2024-10-09 2024-11-22 陕西威尔机电科技有限公司 Texture feature analysis-based three-dimensional contour defect detection method for contact instrument
CN119091054A (en) * 2024-11-06 2024-12-06 中国人民解放军海军航空大学 A method and system for quickly building infrared three-dimensional models
CN119295475A (en) * 2024-12-16 2025-01-10 苏州汇影光学技术有限公司 Photovoltaic cell defect detection method and system based on image processing

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105096268A (en) * 2015-07-13 2015-11-25 西北农林科技大学 Denoising smoothing method of point cloud
CN110473201A (en) * 2019-09-10 2019-11-19 江南大学 A kind of automatic testing method and device of disc surface defect
US20200256999A1 (en) * 2019-02-07 2020-08-13 Atulya Yellepeddi Lidar techniques for autonomous vehicles
EP3739356A1 (en) * 2019-05-12 2020-11-18 Origin Wireless, Inc. Method, apparatus, and system for wireless tracking, scanning and monitoring
CN114186588A (en) * 2021-12-08 2022-03-15 哈尔滨工程大学 A real-time hierarchical filtering method
CN114519676A (en) * 2022-02-15 2022-05-20 上海富瀚微电子股份有限公司 Bayer format-based raw image denoising device and method
CN115222741A (en) * 2022-09-20 2022-10-21 江苏昱恒电气有限公司 Cable surface defect detection method

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105096268A (en) * 2015-07-13 2015-11-25 西北农林科技大学 Denoising smoothing method of point cloud
US20200256999A1 (en) * 2019-02-07 2020-08-13 Atulya Yellepeddi Lidar techniques for autonomous vehicles
EP3739356A1 (en) * 2019-05-12 2020-11-18 Origin Wireless, Inc. Method, apparatus, and system for wireless tracking, scanning and monitoring
CN110473201A (en) * 2019-09-10 2019-11-19 江南大学 A kind of automatic testing method and device of disc surface defect
CN114186588A (en) * 2021-12-08 2022-03-15 哈尔滨工程大学 A real-time hierarchical filtering method
CN114519676A (en) * 2022-02-15 2022-05-20 上海富瀚微电子股份有限公司 Bayer format-based raw image denoising device and method
CN115222741A (en) * 2022-09-20 2022-10-21 江苏昱恒电气有限公司 Cable surface defect detection method

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
LEI CHEN, ET AL: "hierarchical Denoising method of crop 3D point cloud based on multi-view image reconstruction", 《IFIPAICT》, vol. 545 *
赵安安 等: "基于卷积神经网络的航空零件去噪技术", 《机械设计与制造工程》, vol. 2023, no. 01, pages 71 - 74 *

Cited By (40)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116994352A (en) * 2023-08-02 2023-11-03 河北科华防静电地板制造有限公司 Computer lab regularly patrol and examine monitoring system based on intelligence antistatic floor
CN117058670A (en) * 2023-10-12 2023-11-14 深圳市华加生物科技有限公司 Electronic tobacco tar flexibility evaluation method
CN117058670B (en) * 2023-10-12 2024-01-02 深圳市华加生物科技有限公司 Electronic tobacco tar flexibility evaluation method
CN117078667A (en) * 2023-10-13 2023-11-17 山东克莱蒙特新材料科技有限公司 Mineral casting detection method based on machine vision
CN117078667B (en) * 2023-10-13 2024-01-09 山东克莱蒙特新材料科技有限公司 Mineral casting detection method based on machine vision
CN117115494B (en) * 2023-10-23 2024-02-06 卡松科技股份有限公司 Lubricating oil impurity pollution detection method and device based on artificial intelligence
CN117173703B (en) * 2023-11-02 2024-01-16 温州华嘉电器有限公司 Isolating switch state identification method
CN117173703A (en) * 2023-11-02 2023-12-05 温州华嘉电器有限公司 Isolating switch state identification method
CN117292328B (en) * 2023-11-24 2024-02-02 山东新中鲁建设有限公司 Safety management and monitoring method and system for construction quality of assembled building
CN117292328A (en) * 2023-11-24 2023-12-26 山东新中鲁建设有限公司 Safety management and monitoring method and system for construction quality of assembled building
CN117314899A (en) * 2023-11-28 2023-12-29 深圳市烯碳复合材料有限公司 Carbon fiber plate quality detection method based on image characteristics
CN117314899B (en) * 2023-11-28 2024-03-08 深圳市烯碳复合材料有限公司 Carbon fiber plate quality detection method based on image characteristics
CN117392469A (en) * 2023-12-11 2024-01-12 深圳市丰源升科技有限公司 Perovskite battery surface coating detection method and system based on machine vision
CN117392469B (en) * 2023-12-11 2024-02-20 深圳市丰源升科技有限公司 Perovskite battery surface coating detection method and system based on machine vision
CN117426776A (en) * 2023-12-20 2024-01-23 河北雄安九康科技有限公司 Electrocardiogram signal characteristic intelligent extraction method
CN117426776B (en) * 2023-12-20 2024-02-23 河北雄安九康科技有限公司 Electrocardiogram signal characteristic intelligent extraction method
CN117455915A (en) * 2023-12-25 2024-01-26 陕西福坤顺科技有限公司 Machine vision-based three-axis turntable defect detection method
CN117455915B (en) * 2023-12-25 2024-03-08 陕西福坤顺科技有限公司 Machine vision-based three-axis turntable defect detection method
CN117764967A (en) * 2023-12-28 2024-03-26 山西建投集团装饰有限公司 Fitment quality detection method and system based on artificial intelligence
CN117876359A (en) * 2024-03-04 2024-04-12 金品计算机科技(天津)有限公司 Rubber product defect detection method, device, equipment and medium
CN117853493A (en) * 2024-03-08 2024-04-09 山东天意机械股份有限公司 Wallboard production quality detection method based on image processing
CN117853493B (en) * 2024-03-08 2024-05-24 山东天意机械股份有限公司 Wallboard production quality detection method based on image processing
CN117876402A (en) * 2024-03-13 2024-04-12 中国人民解放军总医院第一医学中心 An intelligent segmentation method for temporomandibular joint disorder images
CN117876402B (en) * 2024-03-13 2024-05-07 中国人民解放军总医院第一医学中心 Intelligent segmentation method for temporomandibular joint disorder image
CN117934469A (en) * 2024-03-22 2024-04-26 张家港市升恒机械设备制造有限公司 Visual detection method for production quality of mining high-voltage frequency converter shell
CN117934469B (en) * 2024-03-22 2024-06-07 张家港市升恒机械设备制造有限公司 Visual detection method for production quality of mining high-voltage frequency converter shell
CN117952860A (en) * 2024-03-27 2024-04-30 山东正禾大教育科技有限公司 Mobile digital publishing method and system
CN118014882A (en) * 2024-04-08 2024-05-10 汉中精测电器有限责任公司 Drilling positioning method of torque sensor based on image processing
CN118014882B (en) * 2024-04-08 2024-06-04 汉中精测电器有限责任公司 Drilling positioning method of torque sensor based on image processing
CN118196082B (en) * 2024-05-13 2024-07-26 中铁七局集团第三工程有限公司 Method and system for plugging civil air defense channel
CN118196082A (en) * 2024-05-13 2024-06-14 中铁七局集团第三工程有限公司 Method and system for plugging civil air defense channel
CN118483195A (en) * 2024-05-17 2024-08-13 陕西天域新型环保建材有限公司 Detection method of water retention performance of special gypsum mortar based on optical analysis
CN118823024A (en) * 2024-09-20 2024-10-22 东莞市隆慧电子科技有限公司 A method for identifying machining defects of air-cooled radiators
CN118823024B (en) * 2024-09-20 2024-11-29 东莞市隆慧电子科技有限公司 A method for identifying machining defects of air-cooled radiators
CN118864470A (en) * 2024-09-26 2024-10-29 瓦拉赫(大连)智能科技有限公司 An intelligent analysis system for obstetric ultrasound examination data
CN119006503A (en) * 2024-10-09 2024-11-22 陕西威尔机电科技有限公司 Texture feature analysis-based three-dimensional contour defect detection method for contact instrument
CN119091054A (en) * 2024-11-06 2024-12-06 中国人民解放军海军航空大学 A method and system for quickly building infrared three-dimensional models
CN119091054B (en) * 2024-11-06 2025-06-27 中国人民解放军海军航空大学 Rapid construction method and system of infrared three-dimensional model
CN119295475A (en) * 2024-12-16 2025-01-10 苏州汇影光学技术有限公司 Photovoltaic cell defect detection method and system based on image processing
CN119295475B (en) * 2024-12-16 2025-04-15 苏州汇影光学技术有限公司 Photovoltaic cell defect detection method and system based on image processing

Also Published As

Publication number Publication date
CN116310360B (en) 2023-08-18

Similar Documents

Publication Publication Date Title
CN116310360B (en) Reactor surface defect detection method
CN115457041B (en) Road quality identification and detection method
CN113538433B (en) Mechanical casting defect detection method and system based on artificial intelligence
CN115829883B (en) Surface image denoising method for special-shaped metal structural member
CN117152137B (en) Welded pipe corrosion state detection method based on image processing
CN113963042B (en) Metal part defect degree evaluation method based on image processing
CN105654501B (en) Adaptive Image Segmentation Method Based on Fuzzy Threshold
CN115564767B (en) Inductance winding quality monitoring method based on machine vision
CN116110053B (en) Container surface information detection method based on image recognition
CN115775250A (en) Golden finger circuit board defect rapid detection system based on digital image analysis
CN115330767A (en) Method for identifying production abnormity of corrosion foil
CN115994870B (en) Image processing method for enhancing denoising
CN117788345B (en) Land remote sensing image correction and enhancement method
CN117974989A (en) Rapid detection method for garden plant disease and pest areas
CN117809379B (en) Intelligent humanoid recognition alarm system and method based on monitoring camera
CN117541582B (en) IGBT insulation quality detection method for high-frequency converter
CN117635615B (en) Defect detection method and system for realizing punching die based on deep learning
CN115330758A (en) Welding quality detection method based on denoising processing
CN115511907B (en) Scratch detection method for LED screen
CN119477709B (en) Method for enhancing contour of fluorescence in-situ hybridization image by means of illumination fusion and local binary
CN117974637B (en) Intelligent detection method and system for guide wheel crack defects based on image features
CN117423113B (en) Adaptive denoising method for archive OCR (optical character recognition) image
CN116883401B (en) Industrial product production quality detection system
CN116109658B (en) Harvester control data processing method based on 5G technology
CN116912254A (en) Cable defect identification method based on data enhancement preprocessing

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