CN114820611B - Artificial intelligence-based mechanical part quality evaluation method and system - Google Patents

Artificial intelligence-based mechanical part quality evaluation method and system Download PDF

Info

Publication number
CN114820611B
CN114820611B CN202210744728.XA CN202210744728A CN114820611B CN 114820611 B CN114820611 B CN 114820611B CN 202210744728 A CN202210744728 A CN 202210744728A CN 114820611 B CN114820611 B CN 114820611B
Authority
CN
China
Prior art keywords
distribution
sliding window
image information
mechanical part
artificial intelligence
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
Application number
CN202210744728.XA
Other languages
Chinese (zh)
Other versions
CN114820611A (en
Inventor
徐军
谢昌萍
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.)
Nantong Hengqiang Mill Roll Co ltd
Original Assignee
Nantong Hengqiang Mill Roll 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 Nantong Hengqiang Mill Roll Co ltd filed Critical Nantong Hengqiang Mill Roll Co ltd
Priority to CN202210744728.XA priority Critical patent/CN114820611B/en
Publication of CN114820611A publication Critical patent/CN114820611A/en
Application granted granted Critical
Publication of CN114820611B publication Critical patent/CN114820611B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0004—Industrial image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20092—Interactive image processing based on input by user
    • G06T2207/20104—Interactive definition of region of interest [ROI]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30108—Industrial image inspection
    • G06T2207/30164—Workpiece; Machine component
    • 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
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Landscapes

  • Engineering & Computer Science (AREA)
  • Quality & Reliability (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Image Analysis (AREA)

Abstract

The invention relates to the field of defect detection, in particular to a mechanical part quality evaluation method and a system based on artificial intelligence, wherein the method utilizes an optical means, specifically a visible light means to collect a visible light image so as to test or analyze the mechanical part, specifically to test the existence of flaws or defects of the mechanical part and evaluate the quality of the mechanical part, and specifically comprises the following steps: acquiring a visible light image of the surface of the mechanical part, acquiring an ROI (region of interest) of the visible light image, extracting image information of each channel of the ROI, and acquiring a conversion image corresponding to the image information; acquiring corresponding distribution curved surfaces of each converted image in different sliding window directions to obtain characteristic vectors of the distribution curved surfaces, and calculating the weight of the characteristic vectors to obtain distribution characteristic parameters of image information of each channel; acquiring standard distribution characteristic parameters of a standard mechanical part, and evaluating the quality of the mechanical part; meanwhile, the method can be applied to new material related services, and can realize new material detection, metering, related standardization and the like.

Description

Mechanical part quality evaluation method and system based on artificial intelligence
Technical Field
The invention relates to the technical field of defect detection, in particular to a mechanical part quality evaluation method and system based on artificial intelligence.
Background
In the process of manufacturing and processing mechanical parts, various defects appear on the surfaces of the parts due to the influence of various external factors such as operators, machine equipment, working environment and the like and the self factors of the materials of the parts; for example: the surface is uneven, the flaw is obvious, and the scratch exists; since the mechanical parts are mostly used for bearing a certain pressure or performing a large amount of reciprocating motion in the mechanical system, once the mechanical parts with abnormal defects are used, the whole mechanical system may be crashed; therefore, this step of quality inspection of mechanical parts is very important. Most manufacturers place the part on fixed workstation at present and detect by the manual work, and this kind of detection method not only the cost of labor is high, and personnel intensity of labour is big, and manual detection very easily receives the influence of inspector health, causes to detect the precision lower, appears higher false retrieval rate.
Disclosure of Invention
In order to solve the technical problems, the invention aims to provide a mechanical part quality evaluation method based on artificial intelligence, and the adopted technical scheme is as follows:
acquiring RGB image information containing mechanical parts, and extracting an ROI (region of interest) of the RGB image information, wherein the ROI is a region where the mechanical parts are located;
performing information processing on the ROI to obtain R channel image information, G channel image information and B channel image information corresponding to the ROI; respectively carrying out pixel value conversion on each pixel point in the R channel image information, the G channel image information and the B channel image information to obtain a conversion pixel value corresponding to each pixel point, and further obtaining a conversion image corresponding to each channel image information;
assigning values to the conversion pixel values in sequence according to the size of the conversion pixel values to obtain a plurality of assignment results, and taking the assignment results as elements in a set to obtain a set; randomly selecting one or two elements in the set to form a binary group, wherein when one element in the randomly selected set forms the binary group, the binary group corresponds to two same elements, and when two elements in the randomly selected set form the binary group, the binary group corresponds to two different elements; and performing sliding window operation on one of the converted images in different sliding window directions by using a sliding window with the size of 1 multiplied by 2, acquiring a distribution curved surface corresponding to one of the converted images in the sliding window direction according to the probability of each binary group appearing in the sliding window process corresponding to the sliding window direction, and further acquiring the distribution curved surface corresponding to each converted image in different sliding window directions.
Calculating a first distribution index and a second distribution index corresponding to the distribution curved surface, and obtaining a characteristic vector corresponding to the distribution curved surface according to the first distribution index and the second distribution index;
calculating the weights of the feature vectors of the distribution curved surfaces corresponding to different sliding window directions according to the ROI, and obtaining distribution feature parameters corresponding to image information of each channel according to the weights;
and acquiring standard distribution characteristic parameters corresponding to the standard mechanical parts, and evaluating the quality of the mechanical parts according to the distribution characteristic parameters and the standard distribution characteristic parameters.
Further, the method for obtaining the converted pixel value comprises the following steps: comparing the values of the pixel points with 8 adjacent pixel points, recording the pixel point with the value larger than that of the pixel point as 1, recording the pixel point with the value smaller than that of the pixel point as 0, sequencing the adjacent pixel points from left to right and from top to bottom to form a binary number, and converting the binary number into a decimal number to manufacture a converted pixel value of the pixel point.
Further, the method for obtaining the distribution surface comprises the following steps:
and establishing a three-dimensional coordinate system, wherein the x axis of the three-dimensional coordinate system is an element, the y axis of the three-dimensional coordinate system is an element, and the z axis of the three-dimensional coordinate system is the probability of the occurrence of the binary group, and acquiring the distribution curved surface according to the three-dimensional coordinate system.
Further, the first distribution index is:
Figure 100002_DEST_PATH_IMAGE002
wherein the content of the first and second substances,
Figure 100002_DEST_PATH_IMAGE004
，
Figure 100002_DEST_PATH_IMAGE006
is an element
Figure 100002_DEST_PATH_IMAGE008
And elements
Figure 100002_DEST_PATH_IMAGE010
The two-tuple of the composition is composed of,
Figure 100002_DEST_PATH_IMAGE012
for the set of all the tuples that are made up,
Figure 100002_DEST_PATH_IMAGE014
is a binary group
Figure 100002_DEST_PATH_IMAGE016
Probability of occurrence during sliding window operation.
Further, the second distribution index is:
Figure 100002_DEST_PATH_IMAGE018
wherein the content of the first and second substances,
Figure 100002_DEST_PATH_IMAGE020
，
Figure 94952DEST_PATH_IMAGE006
is an element
Figure 84861DEST_PATH_IMAGE008
And elements
Figure 333440DEST_PATH_IMAGE010
The two-tuple of the composition is composed of,
Figure 72857DEST_PATH_IMAGE012
for the set of all the tuples that are made up,
Figure 356071DEST_PATH_IMAGE014
is a binary group
Figure 522741DEST_PATH_IMAGE016
Probability of occurrence during sliding window operation;
Figure 100002_DEST_PATH_IMAGE022
is composed of
Figure 100002_DEST_PATH_IMAGE024
The average value of (a) of (b),
Figure 100002_DEST_PATH_IMAGE026
is composed of
Figure 100002_DEST_PATH_IMAGE028
The mean value of (a);
in the formula (I), the compound is shown in the specification,
Figure 100002_DEST_PATH_IMAGE030
，
Figure 100002_DEST_PATH_IMAGE032
，
Figure 290843DEST_PATH_IMAGE024
the first element corresponding to a doublet is
Figure 99530DEST_PATH_IMAGE008
The sum of the probabilities corresponding to a second element change while remaining unchanged,
Figure 100002_DEST_PATH_IMAGE034
is the maximum value of the element;
Figure 100002_DEST_PATH_IMAGE036
，
Figure 100002_DEST_PATH_IMAGE038
，
Figure 320471DEST_PATH_IMAGE028
the first element being a change of the binary and the second element being
Figure 189201DEST_PATH_IMAGE010
The sum of the corresponding probabilities when kept constant,
Figure 553318DEST_PATH_IMAGE034
is the maximum value of the element and is,
Figure 100002_DEST_PATH_IMAGE040
is composed of
Figure 775483DEST_PATH_IMAGE024
The variance of (a) is determined,
Figure 100002_DEST_PATH_IMAGE042
is composed of
Figure 643076DEST_PATH_IMAGE028
The variance of (c).
Further, the distribution characteristic parameters are:
Figure 100002_DEST_PATH_IMAGE044
wherein the content of the first and second substances,
Figure 100002_DEST_PATH_IMAGE046
is the direction of the ith sliding window,
Figure 100002_DEST_PATH_IMAGE048
to convert the feature vector of the distribution surface obtained by the image in the ith sliding window direction,
Figure 100002_DEST_PATH_IMAGE050
and m is the total number of the sliding window directions for the weight of the feature vector of the distribution curved surface obtained in the ith sliding window direction of the converted image.
Further, the method for evaluating comprises the following steps: establishing a quality evaluation model according to the distribution characteristic parameters and the standard distribution characteristic parameters, and evaluating the quality of the mechanical part through the quality evaluation model;
the quality evaluation model is as follows:
Figure 100002_DEST_PATH_IMAGE052
wherein the content of the first and second substances,
Figure 100002_DEST_PATH_IMAGE054
as an evaluation value of the image information of the corresponding channel,
Figure 100002_DEST_PATH_IMAGE056
in order to be a standard distribution characteristic parameter,
Figure 100002_DEST_PATH_IMAGE058
in order to distribute the characteristic parameters of the image,
Figure 100002_DEST_PATH_IMAGE060
in order to distribute the modes of the characteristic parameters,
Figure 100002_DEST_PATH_IMAGE062
is the norm of the standard distribution characteristic parameter.
The invention also provides an artificial intelligence based mechanical part quality evaluation system, which comprises a processor and a memory, wherein the processor executes the program of the artificial intelligence based mechanical part quality evaluation method stored in the memory.
The embodiment of the invention at least has the following beneficial effects:
the invention relates to the field of defect detection, in particular to a mechanical part quality evaluation method and a system based on artificial intelligence, wherein the method utilizes an optical means, specifically utilizes a visible light means to collect a visible light image so as to test or analyze the mechanical part, and specifically tests the existence of flaws or defects of the mechanical part to evaluate the quality of the mechanical part; the method takes the randomness of setting in different sliding window directions into consideration, calculates the weights in different sliding window directions through the ROI regional image, performs optimization correction on the feature vectors extracted in different sliding window directions according to the weights to obtain the distribution feature parameters, improves the detection precision of the surface condition of the mechanical part, ensures the accuracy of the extracted distribution feature parameters, and has the advantages of no contact, high detection speed, high evaluation precision and the like. Meanwhile, the method can be applied to new material related services, and can realize new material detection, metering, related standardization and the like.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions and advantages of the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
FIG. 1 is a flow chart illustrating the steps of the method for evaluating the quality of a mechanical part based on artificial intelligence according to the present invention.
Detailed Description
In order to further explain the technical means and effects of the present invention adopted to achieve the predetermined objects, the following detailed description, the structure, the features and the effects of the method and the system for evaluating the quality of a mechanical part based on artificial intelligence according to the present invention are provided with the accompanying drawings and the preferred embodiments. In the following description, different "one embodiment" or "another embodiment" refers to not necessarily the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
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.
Referring to fig. 1, a flowchart illustrating steps of a method for evaluating quality of a mechanical part based on artificial intelligence according to an embodiment of the present invention is shown, the method including the following steps:
step 1, RGB image information containing mechanical parts is obtained, and ROI areas of the RGB image information are extracted, wherein the ROI areas are areas where the mechanical parts are located.
Specifically, RGB image information containing mechanical parts is acquired by a camera; the shooting range and the shooting angle of the camera can be adjusted by an implementer according to actual conditions, so that the shooting range of the camera can cover the surface of the mechanical part to be detected, and the mechanical part can be comprehensively detected; in the embodiment, the camera is deployed above the mechanical part to be detected, so that the surface image of the mechanical part to be detected is acquired at the overlooking visual angle, the surface front-looking image of the mechanical part to be detected is acquired, and the influence of the camera visual angle on the acquisition of the surface image of the mechanical part to be detected is prevented.
Further, in order to reduce the influence of the irrelevant area in the RGB image information on the quality evaluation, the present embodiment extracts the area in which the mechanical part is located in the RGB image information.
The extraction method comprises the following steps: and identifying the mechanical part pixel points in the RGB image information by adopting a semantic segmentation network. The output of the semantic segmentation network is a semantic segmentation effect graph, the semantic segmentation effect graph is a binary image, the pixel value of the pixel point corresponding to the mechanical part in the binary image is 1, and the pixel values of the other remaining pixel points are 0; and multiplying the binary image by the RGB image information to obtain RGB image data corresponding to the mechanical part, namely obtaining the ROI of the RGB image information. The process of segmenting and identifying the image by the semantic segmentation network is a known technology, and is not described herein again.
And 2, processing the information of the ROI to obtain R channel image information, G channel image information and B channel image information corresponding to the ROI, respectively converting pixel values of all pixel points in all the channel image information to obtain converted pixel values corresponding to all the pixel points, and further obtaining converted images corresponding to all the channel image information.
The method for acquiring the converted pixel value comprises the following steps: comparing the values of the pixel points and 8 neighborhood pixel points, recording the pixel point with the value larger than the value of the pixel point as 1, recording the pixel point with the value smaller than the value of the pixel point as 0, sequencing the neighborhood pixel points from left to right and from top to bottom in sequence to form binary numbers, and converting the binary numbers into decimal numbers to manufacture the converted pixel value of the pixel point.
It should be noted that the features of the mechanical part reflected in different channel image information are not the same, and the channel extraction of the ROI is performed to obtain the features of the mechanical part corresponding to the different channel image information, so that the features corresponding to the mechanical part can be obtained more comprehensively and accurately.
The converted image realizes the significance processing of the channel image information, preliminarily obtains the texture distribution information of the pixel points in each channel image information, provides convenience for accurately extracting the distribution condition of the ROI area, and improves the extraction precision of the texture structure characteristics of the ROI area.
Step 3, assigning the conversion pixel values in sequence according to the size of the conversion pixel values to obtain a plurality of assignment results, and taking the assignment results as elements in a set to obtain a set; randomly selecting one or two elements in the set to form a binary group, wherein when one element in the randomly selected set forms the binary group, the binary group corresponds to two same elements, and when two elements in the randomly selected set form the binary group, the binary group corresponds to two different elements; and performing sliding window operation on one of the converted images in different sliding window directions by using a sliding window with the size of 1 multiplied by 2, acquiring a distribution curved surface corresponding to one of the converted images in the sliding window direction according to the probability of each binary group appearing in the sliding window process corresponding to the sliding window direction, and further acquiring the distribution curved surface corresponding to each converted image in different sliding window directions.
In the embodiment, assignment is sequentially performed on the conversion pixel values according to the size of the conversion pixel values to obtain a plurality of assignment results; i.e. the converted pixel values are in the interval 0,15]The value of the inner is 0, and the converted pixel value is in the interval [16,31]The value in the set is 1, and the like, so as to obtain a set, wherein the set is {0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15 }; i.e. 16 elements in the set, then arbitrarily selecting one or two elements in the set to form a binary group
Figure 28502DEST_PATH_IMAGE020
，
Figure DEST_PATH_IMAGE064
The value ranges are all
Figure DEST_PATH_IMAGE066
And the total number of the elements is 16,
Figure 365068DEST_PATH_IMAGE008
is a binary group
Figure 249978DEST_PATH_IMAGE006
The first element of (a);
Figure 627870DEST_PATH_IMAGE010
is a binary group
Figure 307244DEST_PATH_IMAGE006
The second element of (1). In the specific implementation process, an implementer can adjust the number of the elements according to the actual situation.
Setting the step length of a sliding window to be 1, and in the process of performing sliding window operation on the converted image by using the sliding window with the size of 1 multiplied by 2, each sliding window area has a binary group corresponding to the sliding window area, namely the sliding window area can be regarded as a binary group; because each pixel has its corresponding element, the two pixels in the sliding window area can form a binary group.
The method for acquiring the middle distribution curved surface comprises the following steps: and establishing a three-dimensional coordinate system, wherein the x axis of the three-dimensional coordinate system is an element, the x axis corresponds to the first element in the binary group, the value range is 0-15 and totally 16 elements, the y axis of the three-dimensional coordinate system is an element, the value range corresponds to the second element in the binary group, the y axis corresponds to the second element in the binary group and totally 16 elements, the z axis of the three-dimensional coordinate system is the probability of the occurrence of the binary group, and the distribution surface is obtained according to the three-dimensional coordinate system.
The different sliding window directions in this embodiment are: horizontal (including left and right), vertical (including up and down), directions corresponding to the 45 ° line (including 45 ° and-135 °) and directions corresponding to the 135 ° line (including 135 ° and-45 °); the sliding direction of the sliding window refers to the direction of the center line of the short side of the sliding window, and different sliding window directions are set to extract the distribution condition of each pixel point in the converted image under different direction scales, so that the accuracy of extracting the distribution characteristics of each pixel point in the ROI area is ensured.
It should be noted that the distribution condition of the surface of the mechanical part is detected through the binary group, the distribution condition of elements of different combinations on the corresponding conversion image can be detected through the binary group, neighborhood information is considered through the binary group and different sliding window directions, the local distribution condition is represented, the binary group can reflect local texture information of pixel points and neighborhood pixel points thereof, and the distribution condition of each conversion pixel value in the corresponding conversion image is more accurately analyzed.
And 4, calculating a first distribution index and a second distribution index corresponding to the distribution curved surface, and obtaining a characteristic vector corresponding to the distribution curved surface according to the first distribution index and the second distribution index.
The first distribution index is:
Figure DEST_PATH_IMAGE002A
wherein the content of the first and second substances,
Figure 306338DEST_PATH_IMAGE020
，
Figure 463781DEST_PATH_IMAGE006
is an element
Figure 696179DEST_PATH_IMAGE008
And elements
Figure 77613DEST_PATH_IMAGE010
The two-tuple of the composition is composed of,
Figure 372459DEST_PATH_IMAGE012
for the set of all the tuples that are made up,
Figure 723806DEST_PATH_IMAGE014
is a binary group
Figure 420498DEST_PATH_IMAGE016
Probability of occurrence during sliding window operation.
The second distribution index is:
Figure DEST_PATH_IMAGE018A
wherein the content of the first and second substances,
Figure 644937DEST_PATH_IMAGE020
，
Figure 958237DEST_PATH_IMAGE006
is an element
Figure 988642DEST_PATH_IMAGE008
And elements
Figure 930053DEST_PATH_IMAGE010
The two-element group is formed by the following components,
Figure 122131DEST_PATH_IMAGE012
for the set of all the tuples that are made up,
Figure 516203DEST_PATH_IMAGE014
is a binary group
Figure 350298DEST_PATH_IMAGE016
Probability of occurrence during sliding window operation;
Figure 880636DEST_PATH_IMAGE022
is composed of
Figure 509195DEST_PATH_IMAGE024
The average value of (a) is calculated,
Figure 390563DEST_PATH_IMAGE026
is composed of
Figure 774489DEST_PATH_IMAGE028
The mean value of (a);
in the formula (I), the compound is shown in the specification,
Figure 424913DEST_PATH_IMAGE030
，
Figure 958794DEST_PATH_IMAGE032
，
Figure 327458DEST_PATH_IMAGE024
the first element corresponding to a doublet is
Figure 237776DEST_PATH_IMAGE008
The sum of the probabilities corresponding to a second element change while remaining unchanged,
Figure 149232DEST_PATH_IMAGE034
is the maximum value of the element;
Figure 978647DEST_PATH_IMAGE036
，
Figure 709974DEST_PATH_IMAGE038
，
Figure 548617DEST_PATH_IMAGE028
the first element being a change of the binary and the second element being
Figure 49000DEST_PATH_IMAGE010
The sum of the corresponding probabilities when kept constant,
Figure 455841DEST_PATH_IMAGE034
is the maximum value of the element, in this example, n is 15,
Figure 533519DEST_PATH_IMAGE040
is composed of
Figure 316798DEST_PATH_IMAGE024
The variance of (a) is calculated,
Figure 796321DEST_PATH_IMAGE042
is composed of
Figure 108485DEST_PATH_IMAGE028
The variance of (c).
The feature vector in the above is
Figure DEST_PATH_IMAGE068
，
Figure DEST_PATH_IMAGE070
In order to convert the first distribution index of the image corresponding to the distribution surface in the ith sliding window direction,
Figure DEST_PATH_IMAGE072
a second distribution index of the corresponding distribution curved surface of the converted image in the ith sliding window direction; wherein the content of the first and second substances,
Figure 96295DEST_PATH_IMAGE046
for the direction of the ith sliding window,
Figure DEST_PATH_IMAGE074
，
Figure DEST_PATH_IMAGE076
for the 1 st sliding window direction to correspond to the horizontal direction (including left and right) in the present embodiment,
Figure DEST_PATH_IMAGE078
the 2 nd sliding window direction corresponds to the vertical direction (including up and down) in the present embodiment,
Figure DEST_PATH_IMAGE080
the 3 rd sliding window direction corresponds to the direction (including 45 degrees and-135 degrees) corresponding to the straight line of 45 degrees in the embodiment,
Figure DEST_PATH_IMAGE082
the 4 th sliding window direction corresponds to the direction (including 135 ° and-45 °) corresponding to the line of 135 ° in the present embodiment.
It should be noted that the first distribution index represents the complexity of the ROI, and the higher the complexity, the higher the value of the corresponding first distribution index, and the second distribution index represents the correlation degree of the distribution surface in the row-column direction.
And 5, calculating the weights of the feature vectors of the distribution curved surfaces corresponding to different sliding window directions according to the ROI, and acquiring the distribution feature parameters corresponding to the image information of each channel by using the weights.
The weight is:
Figure DEST_PATH_IMAGE084
wherein the content of the first and second substances,
Figure 782798DEST_PATH_IMAGE046
for the direction of the ith sliding window,
Figure 257773DEST_PATH_IMAGE050
for converting the characteristic vectors of the distribution surface obtained in the direction of the ith sliding windowThe weight of the weight is calculated,
Figure DEST_PATH_IMAGE086
and m is the distribution condition index corresponding to the ith sliding window direction, and the total number of the sliding window directions. The total number of sliding window directions in this embodiment is 4.
When i is 1, it corresponds to the horizontal direction (including left and right) in the present embodiment.
Figure DEST_PATH_IMAGE088
Wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE090
as coordinates in the ROI area of
Figure DEST_PATH_IMAGE092
The gray value of the pixel point is set,
Figure DEST_PATH_IMAGE094
as coordinates in the ROI area of
Figure DEST_PATH_IMAGE096
The gray value of the pixel point is set,
Figure DEST_PATH_IMAGE098
as coordinates in the ROI area of
Figure DEST_PATH_IMAGE100
The gray value of the pixel point is set,
Figure DEST_PATH_IMAGE102
as coordinates in the ROI area of
Figure DEST_PATH_IMAGE104
The gray value of the pixel point is set,
Figure DEST_PATH_IMAGE106
as coordinates in the ROI area of
Figure 896432DEST_PATH_IMAGE106
Gray value of pixel point, image size of ROI is
Figure DEST_PATH_IMAGE108
。
When i is 2, it corresponds to the vertical direction (including up and down) in the present embodiment.
Figure DEST_PATH_IMAGE110
Wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE112
as coordinates in the ROI area of
Figure DEST_PATH_IMAGE114
The gray value of the pixel point is set,
Figure DEST_PATH_IMAGE116
as coordinates in the ROI area of
Figure DEST_PATH_IMAGE118
The gray value of the pixel point is set,
Figure 312151DEST_PATH_IMAGE098
as coordinates in the ROI area of
Figure 30708DEST_PATH_IMAGE100
The gray value of the pixel point is set,
Figure DEST_PATH_IMAGE120
as coordinates in the ROI area of
Figure DEST_PATH_IMAGE122
The gray value of the pixel point is set,
Figure DEST_PATH_IMAGE124
as coordinates in the ROI area of
Figure 907659DEST_PATH_IMAGE124
Gray value of pixel point, image size of ROI is
Figure 561626DEST_PATH_IMAGE108
。
When i is 3, it corresponds to the direction (including 45 ° and-135 °) corresponding to the line of 45 ° in the present embodiment.
Figure DEST_PATH_IMAGE126
Wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE128
as coordinates in the ROI area of
Figure DEST_PATH_IMAGE130
The gray value of the pixel point is set,
Figure DEST_PATH_IMAGE132
as coordinates in the ROI area of
Figure DEST_PATH_IMAGE134
The gray value of the pixel point is set,
Figure 200724DEST_PATH_IMAGE098
as coordinates in the ROI area of
Figure 457393DEST_PATH_IMAGE100
The gray value of the pixel point is set,
Figure DEST_PATH_IMAGE136
as coordinates in the ROI area of
Figure DEST_PATH_IMAGE138
The gray value of the pixel point is set,
Figure DEST_PATH_IMAGE140
as coordinates in the ROI area of
Figure DEST_PATH_IMAGE142
Gray value of pixel point, image size of ROI is
Figure 267480DEST_PATH_IMAGE108
。
When i is 4, it corresponds to the direction (including 135 ° and-45 °) corresponding to the line of 135 ° in the present embodiment.
Figure DEST_PATH_IMAGE144
Wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE146
as coordinates in the ROI area of
Figure DEST_PATH_IMAGE148
The gray value of the pixel point is set,
Figure DEST_PATH_IMAGE150
as coordinates in the ROI area of
Figure DEST_PATH_IMAGE152
The gray value of the pixel point is set,
Figure 62654DEST_PATH_IMAGE098
as coordinates in the ROI area of
Figure 976601DEST_PATH_IMAGE100
The gray value of the pixel point is set,
Figure DEST_PATH_IMAGE154
as coordinates in the ROI area of
Figure DEST_PATH_IMAGE156
The gray value of the pixel point is set,
Figure DEST_PATH_IMAGE158
as coordinates in the ROI area of
Figure DEST_PATH_IMAGE160
Gray value of pixel point, image size of ROI is
Figure 335163DEST_PATH_IMAGE108
。
The distribution characteristic parameters in the above are:
Figure DEST_PATH_IMAGE044A
wherein the content of the first and second substances,
Figure 983444DEST_PATH_IMAGE046
is the direction of the ith sliding window,
Figure 103847DEST_PATH_IMAGE048
to convert the feature vector of the distribution surface obtained by the image in the ith sliding window direction,
Figure 758951DEST_PATH_IMAGE050
and m is the total number of the sliding window directions for converting the weight of the feature vector of the distribution curved surface obtained by the image in the ith sliding window direction. In step 4, 4 sliding window directions are mentioned, so m in this embodiment is 4.
It should be noted that, in the embodiment, it is considered that, when the sliding direction is set, the sliding direction is not limited according to the trend of the texture information of the ROI area image, the texture feature of the ROI area image is weakened to a certain extent, and the problem of inaccurate extraction of the texture of the edge structure exists, which further causes a certain difference between the obtained feature vector and the feature vector of the ROI area image, so that, in order to improve the extraction precision of the feature vector and achieve the accuracy of extraction of the surface structure distribution feature of the mechanical part, the embodiment determines the texture information in different sliding window directions by using the gradient information of the ROI area image, and optimizes the distribution feature parameter of the surface of the mechanical part according to the feature basis of the ROI area image.
And 6, evaluating the quality of the mechanical part according to the distribution characteristic parameters and the standard distribution characteristic parameters corresponding to the standard mechanical part.
Specifically, the method in step 1 to step 5 is used to obtain the standard distribution characteristic parameters corresponding to the image information of each channel corresponding to the standard mechanical part, and the specific implementation method in step 1 to step 5 has been described in detail in the above-mentioned content, and thus is not described in detail. And establishing a quality evaluation model of the corresponding channel image information according to the distribution characteristic parameters and the corresponding standard distribution characteristic parameters.
The quality evaluation model is as follows:
Figure DEST_PATH_IMAGE052A
wherein the content of the first and second substances,
Figure 170471DEST_PATH_IMAGE054
as an evaluation value of the image information of the corresponding channel,
Figure 486309DEST_PATH_IMAGE056
in order to be a standard distribution characteristic parameter,
Figure 308771DEST_PATH_IMAGE058
in order to distribute the characteristic parameters of the device,
Figure 190184DEST_PATH_IMAGE060
in order to distribute the modes of the characteristic parameters,
Figure 998871DEST_PATH_IMAGE062
is the norm of the standard distribution characteristic parameter.
In this embodiment, each channel image information has a corresponding evaluation value, and a mean value of the evaluation values corresponding to each channel image information is used as an evaluation index of the mechanical part, where the higher the evaluation index is, the higher the quality of the corresponding mechanical part is, the less the defect condition of the surface of the mechanical part is, and the lower the evaluation index is, the lower the quality of the corresponding mechanical part is, and the more the defect condition of the surface of the mechanical part is.
Further, in order to facilitate visual judgment of the quality of the mechanical part by related workers, a threshold value is set, the quality of the mechanical part corresponding to the evaluation index larger than the threshold value is judged to be excellent, the quality of the mechanical part corresponding to the average value smaller than the threshold value is judged to be poor, the mechanical part with the excellent quality meets the requirements of delivery and subsequent use, the mechanical part with the poor quality has serious surface defects, and needs to be processed again to meet the delivery requirement, so that accidents in the subsequent use process are prevented.
The invention also provides a mechanical part quality evaluation system based on artificial intelligence, which comprises a processor and a memory, wherein the processor executes the program of the mechanical part quality evaluation method based on artificial intelligence stored in the memory. Since the specific implementation of the method for evaluating the quality of a mechanical part based on artificial intelligence is given in detail in the above steps 1 to 6, redundant description is omitted.
The above-mentioned embodiments are only used for illustrating the technical solutions of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not substantially depart from the spirit and scope of the embodiments of the present application and are intended to be included within the scope of the present application.

Claims (8)

1. The method for evaluating the quality of the mechanical part based on artificial intelligence is characterized by comprising the following steps of:
acquiring RGB image information containing mechanical parts, and extracting an ROI (region of interest) of the RGB image information, wherein the ROI is a region where the mechanical parts are located;
performing information processing on the ROI to obtain R channel image information, G channel image information and B channel image information corresponding to the ROI; respectively carrying out pixel value conversion on each pixel point in the R channel image information, the G channel image information and the B channel image information to obtain a conversion pixel value corresponding to each pixel point, and further obtaining a conversion image corresponding to each channel image information;
assigning values to the conversion pixel values in sequence according to the size of the conversion pixel values to obtain a plurality of assignment results, and taking the assignment results as elements in a set to obtain a set; randomly selecting one or two elements in the set to form a binary group, wherein when one element in the randomly selected set forms the binary group, the binary group corresponds to two same elements, and when two elements in the randomly selected set form the binary group, the binary group corresponds to two different elements; performing sliding window operation on one of the converted images in different sliding window directions by using a sliding window with the size of 1 multiplied by 2, acquiring a distribution curved surface corresponding to one of the converted images in the sliding window direction according to the probability of each binary group appearing in the sliding window process corresponding to the sliding window direction, and further acquiring the distribution curved surface corresponding to each converted image in different sliding window directions;
calculating a first distribution index and a second distribution index corresponding to the distribution curved surface, and obtaining a feature vector corresponding to the distribution curved surface according to the first distribution index and the second distribution index;
calculating the weights of the feature vectors of the distribution curved surfaces corresponding to different sliding window directions according to the ROI, and acquiring distribution feature parameters corresponding to image information of each channel by using the weights;
and acquiring standard distribution characteristic parameters corresponding to the standard mechanical parts, and evaluating the quality of the mechanical parts according to the distribution characteristic parameters and the standard distribution characteristic parameters.
2. The artificial intelligence based machine part quality assessment method according to claim 1, wherein said converted pixel values are obtained by: comparing the values of the pixel points with 8 adjacent pixel points, recording the pixel point with the value larger than the pixel point as 1, recording the pixel point with the value smaller than the pixel point as 0, sequencing the adjacent pixel points in sequence from left to right from top to bottom to form binary numbers, and converting the binary numbers into decimal numbers to manufacture the converted pixel value of the pixel point.
3. The artificial intelligence based machine part quality evaluation method of claim 1,
the acquisition method of the distribution curved surface comprises the following steps:
and establishing a three-dimensional coordinate system, wherein the x axis of the three-dimensional coordinate system is an element, the y axis of the three-dimensional coordinate system is an element, and the z axis of the three-dimensional coordinate system is the probability of the occurrence of the binary group, and acquiring the distribution curved surface according to the three-dimensional coordinate system.
4. The artificial intelligence based machine part quality evaluation method of claim 1,
the first distribution index is:
Figure DEST_PATH_IMAGE002
wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE004
，
Figure DEST_PATH_IMAGE006
is an element
Figure DEST_PATH_IMAGE008
And elements
Figure DEST_PATH_IMAGE010
The two-element group is formed by the following components,
Figure DEST_PATH_IMAGE012
for the set of all the tuples that are made up,
Figure DEST_PATH_IMAGE014
is a binary group
Figure DEST_PATH_IMAGE016
Probability of occurrence during a sliding window operation.
5. The artificial intelligence based machine part quality evaluation method of claim 1,
the second distribution index is:
Figure DEST_PATH_IMAGE018
wherein, the first and the second end of the pipe are connected with each other,
Figure DEST_PATH_IMAGE020
，
Figure 165919DEST_PATH_IMAGE006
is an element
Figure 503491DEST_PATH_IMAGE008
And elements
Figure 133186DEST_PATH_IMAGE010
The two-tuple of the composition is composed of,
Figure 999425DEST_PATH_IMAGE012
for the set of all the tuples that are made up,
Figure 398176DEST_PATH_IMAGE014
is a binary group
Figure 31283DEST_PATH_IMAGE016
Probability of occurrence during sliding window operation;
Figure DEST_PATH_IMAGE022
is composed of
Figure DEST_PATH_IMAGE024
The average value of (a) of (b),
Figure DEST_PATH_IMAGE026
is composed of
Figure DEST_PATH_IMAGE028
The mean value of (a);
in the formula (I), the compound is shown in the specification,
Figure DEST_PATH_IMAGE030
，
Figure DEST_PATH_IMAGE032
，
Figure 462789DEST_PATH_IMAGE024
the first element corresponding to a doublet is
Figure 878858DEST_PATH_IMAGE008
The sum of the probabilities corresponding to a second element change while remaining unchanged,
Figure DEST_PATH_IMAGE034
is the maximum value of the element;
Figure DEST_PATH_IMAGE036
，
Figure DEST_PATH_IMAGE038
，
Figure 677922DEST_PATH_IMAGE028
the first element being a change of the binary and the second element being
Figure 888455DEST_PATH_IMAGE010
The sum of the corresponding probabilities when kept constant,
Figure 289481DEST_PATH_IMAGE034
is the maximum value of the element and is,
Figure DEST_PATH_IMAGE040
is composed of
Figure 119027DEST_PATH_IMAGE024
The variance of (a) is calculated,
Figure DEST_PATH_IMAGE042
is composed of
Figure 571000DEST_PATH_IMAGE028
The variance of (c).
6. The artificial intelligence based machine part quality assessment method according to claim 1,
the distribution characteristic parameters are as follows:
Figure DEST_PATH_IMAGE044
wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE046
for the direction of the ith sliding window,
Figure DEST_PATH_IMAGE048
in order to convert the feature vector of the distribution surface obtained by the image in the ith sliding window direction,
Figure DEST_PATH_IMAGE050
and m is the total number of the sliding window directions for the weight of the feature vector of the distribution curved surface obtained in the ith sliding window direction of the converted image.
7. The artificial intelligence based machine part quality assessment method according to claim 1 or 6, wherein said assessment method is: establishing a quality evaluation model according to the distribution characteristic parameters and the standard distribution characteristic parameters, and evaluating the quality of the mechanical part through the quality evaluation model;
the quality evaluation model is as follows:
Figure DEST_PATH_IMAGE052
wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE054
as an evaluation value of the image information of the corresponding channel,
Figure DEST_PATH_IMAGE056
is a standard distribution characteristic parameter of the optical fiber,
Figure DEST_PATH_IMAGE058
in order to distribute the characteristic parameters of the device,
Figure DEST_PATH_IMAGE060
in order to distribute the modes of the characteristic parameters,
Figure DEST_PATH_IMAGE062
is the norm of the standard distribution characteristic parameter.
8. An artificial intelligence based quality assessment system for mechanical parts comprising a processor and a memory, characterized in that said processor executes a program of the artificial intelligence based quality assessment method for mechanical parts according to any one of claims 1-7 as stored in said memory.
CN202210744728.XA 2022-06-29 2022-06-29 Artificial intelligence-based mechanical part quality evaluation method and system Active CN114820611B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210744728.XA CN114820611B (en) 2022-06-29 2022-06-29 Artificial intelligence-based mechanical part quality evaluation method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210744728.XA CN114820611B (en) 2022-06-29 2022-06-29 Artificial intelligence-based mechanical part quality evaluation method and system

Publications (2)

Publication Number Publication Date
CN114820611A CN114820611A (en) 2022-07-29
CN114820611B true CN114820611B (en) 2022-09-02

Family

ID=82522327

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210744728.XA Active CN114820611B (en) 2022-06-29 2022-06-29 Artificial intelligence-based mechanical part quality evaluation method and system

Country Status (1)

Country Link
CN (1) CN114820611B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114972354B (en) * 2022-08-02 2022-10-28 济宁金筑新型建材科技有限公司 Image processing-based autoclaved aerated concrete block production control method and system

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
RU2510043C1 (en) * 2012-11-12 2014-03-20 Российская Федерация, от имени которой выступает Государственная корпорация по атомной энергии "Росатом" Method of determining range to earth's surface
CN106291502B (en) * 2016-08-30 2018-11-09 北京航空航天大学 The maximum probability time-domain processing method of background extracting and counteracting in target rcs measurement
CN107123109A (en) * 2017-01-13 2017-09-01 陕西师范大学 A kind of window sliding algorithm detected for Bridge Crack
CN113326855B (en) * 2021-06-22 2022-07-12 长光卫星技术股份有限公司 Night lamplight remote sensing image feature extraction method based on two-dimensional Gaussian surface fitting

Also Published As

Publication number Publication date
CN114820611A (en) 2022-07-29

Similar Documents

Publication Publication Date Title
CN115170576B (en) Aluminum pipe surface defect detection method based on machine vision
CN114549522B (en) Textile quality detection method based on target detection
CN115082418B (en) Precise identification method for automobile parts
CN110148130B (en) Method and device for detecting part defects
CN115239704B (en) Accurate detection and repair method for wood surface defects
CN112651968B (en) Wood board deformation and pit detection method based on depth information
CN108846397B (en) Automatic detection method for cable semi-conducting layer based on image processing
CN111915572B (en) Adaptive gear pitting quantitative detection system and method based on deep learning
CN114943739A (en) Aluminum pipe quality detection method
CN114170208A (en) Paper product defect detection method based on artificial intelligence
CN116030060B (en) Plastic particle quality detection method
CN115100206B (en) Printing defect identification method for textile with periodic pattern
CN109859160A (en) Almag internal defect in cast image-recognizing method based on machine vision
CN113538424B (en) Wood board joint classification data identification method based on artificial intelligence
CN114820611B (en) Artificial intelligence-based mechanical part quality evaluation method and system
CN113313107B (en) Intelligent detection and identification method for multiple types of diseases on cable surface of cable-stayed bridge
CN113155839A (en) Steel plate outer surface defect online detection method based on machine vision
CN113177924A (en) Industrial production line product flaw detection method
CN111724376B (en) Paper disease detection method based on texture feature analysis
CN114581805A (en) Coating roller surface defect detection method adopting 3D line laser profile technology
CN114549441A (en) Sucker defect detection method based on image processing
CN110544243B (en) Automatic detection, quantification and reliability evaluation method for small defects of CT (computed tomography) image
CN113899349A (en) Sea wave parameter detection method, equipment and storage medium
CN114155226A (en) Micro defect edge calculation method
CN113362276A (en) Visual detection method and system for plate

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