CN115690108B - Aluminum alloy rod production quality assessment method based on image processing - Google Patents
Aluminum alloy rod production quality assessment method based on image processing Download PDFInfo
- Publication number
- CN115690108B CN115690108B CN202310005137.5A CN202310005137A CN115690108B CN 115690108 B CN115690108 B CN 115690108B CN 202310005137 A CN202310005137 A CN 202310005137A CN 115690108 B CN115690108 B CN 115690108B
- Authority
- CN
- China
- Prior art keywords
- aggregation
- blocks
- gradient
- determining
- index
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 229910000838 Al alloy Inorganic materials 0.000 title claims abstract description 44
- 238000012545 processing Methods 0.000 title claims abstract description 26
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 23
- 238000001303 quality assessment method Methods 0.000 title abstract description 10
- 238000004220 aggregation Methods 0.000 claims abstract description 264
- 230000002776 aggregation Effects 0.000 claims abstract description 264
- 230000007547 defect Effects 0.000 claims abstract description 215
- 238000000034 method Methods 0.000 claims abstract description 21
- 238000013441 quality evaluation Methods 0.000 claims abstract description 16
- 229910052751 metal Inorganic materials 0.000 claims abstract description 12
- 239000002184 metal Substances 0.000 claims abstract description 12
- 238000004364 calculation method Methods 0.000 claims description 8
- 230000002950 deficient Effects 0.000 claims description 6
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 230000008859 change Effects 0.000 description 13
- 238000005242 forging Methods 0.000 description 10
- 239000011148 porous material Substances 0.000 description 10
- 230000008569 process Effects 0.000 description 7
- 230000011218 segmentation Effects 0.000 description 6
- 238000001514 detection method Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 230000003313 weakening effect Effects 0.000 description 3
- 229910000831 Steel Inorganic materials 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 2
- 230000004931 aggregating effect Effects 0.000 description 2
- 238000005266 casting Methods 0.000 description 2
- 238000002425 crystallisation Methods 0.000 description 2
- 230000008025 crystallization Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 239000010959 steel Substances 0.000 description 2
- 241001270131 Agaricus moelleri Species 0.000 description 1
- 229910052782 aluminium Inorganic materials 0.000 description 1
- XAGFODPZIPBFFR-UHFFFAOYSA-N aluminium Chemical compound [Al] XAGFODPZIPBFFR-UHFFFAOYSA-N 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 238000007670 refining Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000035939 shock Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- 238000011179 visual inspection Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- 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
Abstract
The invention relates to the technical field of image processing, in particular to an aluminum alloy rod production quality evaluation method based on image processing, which comprises the steps of obtaining a surface image of a metal piece, and performing image processing on the surface image to obtain a gray image; determining each target defect pixel point in the gray level image, and further obtaining each aggregation block; combining the distance between any two aggregation blocks and the gray value of each pixel point and the surrounding neighborhood pixel points on the connecting line to determine each candidate cluster radius, and finally determining the optimal density cluster radius; and performing density clustering on the gray level image by using the optimal density clustering radius to obtain a defect area in the gray level image, and further determining a quality evaluation result. According to the invention, the surface image of the metal piece is subjected to corresponding image processing, and the optimal density clustering radius is determined in a self-adaptive manner, so that the defect area in the image can be accurately divided, and finally the accuracy of metal piece production quality assessment is improved.
Description
Technical Field
The invention relates to the technical field of image data processing, in particular to an aluminum alloy rod production quality assessment method based on image processing.
Background
The aluminum alloy forging has the characteristics of high precision, good shock resistance, good corrosion resistance and the like, and the air hole defect often occurs in the production process due to improper operation or unskilled process. For example, in the production process of aluminum alloy rods, a small amount of air holes are occasionally formed in the middle part of the poured aluminum alloy rods, and the air holes are formed due to unreasonable design of pouring caps or poor refining effect of aluminum liquid, so that water seeps on the surfaces of crystallization wheels and steel belts, or oil volatilizes on the surfaces of the crystallization wheels and the steel belts during casting. When the aluminum alloy forging piece has the air hole defect, the casting is often scrapped due to the quality problem, so that the production cost is increased, and therefore, the surface air hole defect of the produced aluminum alloy forging piece needs to be detected.
The conventional detection of the air hole defect of the aluminum alloy forging is generally realized by manual visual inspection, but the detection result has poor reliability and wastes a large amount of manpower. With the development of computer machine vision technology, the surface defect detection method based on image processing is gradually applied to the detection of the air hole defect of the aluminum alloy forging. For example, in the prior art, by acquiring a surface gray image of an aluminum alloy forging piece, and clustering the pixels in the gray image by using a DBSCAN density clustering algorithm based on gray levels and pixel positions of the pixels in the gray image, the pixels belonging to the same region are classified, so that each air hole region can be determined. However, when the DBSCAN density clustering algorithm is adopted for clustering, the clustering radius needs to be set manually, the accuracy of the clustering radius cannot be guaranteed, when the clustering radius is large, under-segmentation occurs, pixel points in different areas are gathered into one type, when the clustering radius is too small, excessive segmentation is caused, the same area is gathered into a plurality of types, and finally the determined air hole areas are inaccurate, so that the quality evaluation result of the aluminum alloy forged piece is affected.
Disclosure of Invention
The invention aims to provide an aluminum alloy rod production quality assessment method based on image processing, which is used for solving the problem that the existing artificial determination of cluster radius is used for identifying air holes on the surface of an aluminum alloy forging piece, and the identification result is inaccurate, so that the quality assessment result of the aluminum alloy forging piece is unreliable.
In order to solve the technical problems, the invention provides an aluminum alloy rod production quality evaluation method based on image processing, which comprises the following steps:
acquiring a surface image of a metal piece to be evaluated, and performing image processing on the surface image to obtain a gray level image;
determining each target defect pixel point according to the gray value of each pixel point in the gray image, and carrying out aggregation grouping on each target defect pixel point according to the position of each target defect pixel point to obtain each aggregation block and the central defect point of each aggregation block;
according to the gray values of each pixel point and the neighborhood pixel points around the pixel points on the central defect point connecting line of every two aggregation blocks, determining the gradient mean value corresponding to each pixel point on the central defect point connecting line of every two aggregation blocks;
determining the relevance index of each two aggregation blocks according to the gradient mean value corresponding to each pixel point on the central defect point connecting line of each two aggregation blocks and the distance value between the central defect points of each two aggregation blocks;
Determining the aggregation index of each two aggregation blocks according to the gradient mean value corresponding to each pixel point on the central defect point connecting line of each two aggregation blocks and the association index of each two aggregation blocks;
determining each candidate clustering radius according to the aggregation index of each two aggregation blocks, the gray value of each pixel point on the central defect point connecting line of each two aggregation blocks and the gradient mean value corresponding to each pixel point;
and determining an optimal density clustering radius according to each candidate clustering radius, and performing density clustering on each pixel point in the gray level image according to the optimal density clustering radius, so as to obtain a defect area in the gray level image, and further determining a quality evaluation result of the metal piece.
Further, determining each target defect pixel point includes:
determining a gray level histogram according to the gray level value of each pixel point in the gray level image, and determining a gray level curve according to the gray level histogram;
determining target peaks in the peaks on the gray scale curves, and determining maximum gray scale values corresponding to the target peaks;
and determining each target defect pixel point in the gray level image according to the maximum gray level value corresponding to the target peak, wherein the gray level value of the target defect pixel point is equal to the maximum gray level value.
Further, determining a gradient mean value corresponding to each pixel point on the central defect point connecting line of every two aggregation blocks includes:
according to the gray values of each pixel point and the neighborhood pixel points around the pixel points on the central defect point connecting line of every two aggregation blocks, determining the gray gradient of each pixel point and the neighborhood pixel points around the pixel points on the central defect point connecting line of every two aggregation blocks;
and constructing a sliding window area of each pixel point on the connecting line of the central defect points of every two aggregation blocks, and determining the average value of gray gradient of each pixel point in the sliding window area as the average value of gradient corresponding to the pixel point on the connecting line corresponding to the sliding window area.
Further, determining the relevance index of each two aggregation blocks includes:
sequencing the gradient mean values corresponding to the pixel points on the central defect point connecting lines of every two aggregation blocks according to the direction from one end to the other end of the central defect point connecting lines of every two aggregation blocks, so as to obtain a gradient mean value sequence of every two aggregation blocks;
determining a difference value between any two adjacent gradient mean values in the gradient mean value sequence of every two aggregation blocks, determining a target gradient mean value according to two gradient mean values corresponding to the difference value between the largest two adjacent gradient mean values, and determining a relevance index of every two aggregation blocks according to the difference value between the largest two adjacent gradient mean values corresponding to every two aggregation blocks, the target gradient mean value and the distance value between the center defect points of every two aggregation blocks, wherein the relevance index and the difference value between the corresponding largest two adjacent gradient mean values, the target gradient mean value and the distance value are all in a negative correlation relationship.
Further, a calculation formula corresponding to the relevance index of each two aggregation blocks is determined as follows:
wherein ,
for the relevance index of every two aggregation blocks,
for the difference between the largest two adjacent gradient averages corresponding to each two aggregate blocks,
for each two corresponding target gradient averages of the aggregate block,
for the distance value between the center defect points of every two aggregation blocks,expto be with natural constanteIs an exponential function of the base.
Further, determining an aggregation indicator for each two aggregation blocks includes:
for the gradient mean value sequence of every two aggregation blocks, determining the maximum gradient mean value in the gradient mean value sequence, calculating the mean value of the difference values between the maximum gradient mean value and each gradient mean value in front of the gradient mean value sequence, thereby obtaining a first gradient difference index, and calculating the mean value of the difference values between the maximum gradient mean value and each gradient mean value behind the gradient mean value, thereby obtaining a second gradient difference index;
and determining the aggregation index of each two aggregation blocks according to the corresponding correlation index, the first gradient difference index and the second gradient difference index of each two aggregation blocks, wherein the absolute difference value and the correlation index of the aggregation index and the corresponding first gradient difference index and second gradient difference index are positive correlation.
Further, the calculation formula corresponding to the aggregation index of each two aggregation blocks is determined as follows:
wherein ,
for the aggregate index of every two aggregate blocks,
for the first gradient difference index corresponding to every two aggregation blocks,
for the second gradient difference index corresponding to every two aggregation blocks, e is a natural constant,expto be with natural constanteK is the relevance index of every two aggregation blocks as an exponential function of the base.
Further, determining each candidate cluster radius includes:
according to the gray values of all the pixel points on the connecting line of the central defect points of every two aggregation blocks and the gradient average values corresponding to all the pixel points, determining the maximum value in the gradient average values corresponding to all the pixel points on the connecting line of the central defect points of every two aggregation blocks, and determining the gray value of the pixel point on the connecting line corresponding to the maximum value in the gradient average values as the edge gray value of every two aggregation blocks;
and determining candidate cluster radiuses corresponding to every two aggregation blocks according to the gray values of the center defect points of every two aggregation blocks, the edge gray values of every two aggregation blocks and the aggregation index, wherein the difference values of the candidate cluster radiuses and the gray values of the corresponding center defect points and the edge gray values are positive correlation relations, and the candidate cluster radiuses and the corresponding aggregation index are negative correlation relations.
Further, determining an optimal density cluster radius includes:
and determining the smallest candidate cluster radius in the candidate cluster radii as the optimal density cluster radius.
Further, the central defect point of each aggregation block is a clustering central point corresponding to each aggregation block obtained by aggregating and grouping each target defect pixel point.
The invention has the following beneficial effects: the surface image of the metal piece to be evaluated is acquired, and is processed in an image processing mode, so that a gray level image can be obtained. According to the change relation of the pixel gray in the gray image, each target defect pixel in the gray image can be accurately determined. Because the target defect pixel points are abnormal pixel points, each aggregation block can be accurately clustered according to the positions of the target defect pixel points through aggregation grouping, namely clustering, and each aggregation block corresponds to one suspected defect area. By considering the gradient mean value corresponding to each pixel point on the connecting line of every two aggregation blocks, the edge distance condition of the two aggregation blocks corresponding to two suspected defect areas can be accurately estimated, and the relevance index of every two aggregation blocks is determined by combining the distance value between every two aggregation blocks, and the relevance index can accurately represent the distance between the two aggregation blocks corresponding to two suspected defect areas. Because the edge position relationship of the two suspected defect areas also influences the selection of the clustering radius, the situation of the edge position relationship of the two suspected defect areas corresponding to the two clustered blocks is evaluated according to the gradient mean value corresponding to each pixel point on the central defect point connecting line of each two clustered blocks, for example, whether the edges of the two suspected defect areas are far away or are bordered, and the relevance index is combined to determine the relevance index of each two clustered blocks, wherein the relevance index comprehensively considers the distance and the edge position relationship of the two suspected defect areas, and can accurately represent the position relationship of the two suspected defect areas. The difference of the edge gray level and the center gray level of the suspected defect area can also influence the selection of the clustering radius, so that the difference of the edge gray level and the center gray level of the suspected defect area can be analyzed according to the gray level value of each pixel point on the connecting line of the center defect points of every two aggregation blocks and the gradient average value corresponding to each pixel point, and the candidate clustering radius corresponding to every two aggregation blocks can be accurately determined by combining with the aggregation index. And screening the candidate cluster radiuses to obtain the optimal density cluster radiuses suitable for all suspected defect areas. Because the optimal density clustering radius is matched with the gray level image corresponding to the surface image of the metal piece, when density clustering algorithm is adopted to perform density clustering on each pixel point in the gray level image, the defect area in the gray level image can be accurately divided, and therefore the accuracy of metal piece production quality assessment is finally improved.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of an image processing-based aluminum alloy rod production quality assessment method of the present invention;
FIG. 2 is an image schematic of a portion of an RGB image of the surface of an aluminum alloy rod of the present invention;
fig. 3 is a schematic diagram of the positional relationship between suspected air hole defect regions where different aggregation blocks are located according to the present invention.
Detailed Description
In order to further describe the technical means and effects adopted by the present invention to achieve the preset purpose, the following detailed description is given below of the specific implementation, structure, features and effects of the technical solution according to the present invention 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.
Because different clustering radiuses are selected to cause different segmentation results of images when a DBSCAN density clustering algorithm is adopted to carry out clustering segmentation on pixel points in the images, the problem that the existing artificial determination of the clustering radiuses is used for identifying air holes on the surface of an aluminum alloy forging part is inaccurate and further the quality evaluation result of the aluminum alloy forging part is unreliable is solved.
Specifically, a flow chart corresponding to the image processing-based aluminum alloy rod production quality evaluation method is shown in fig. 1, and the method comprises the following steps:
step S1: and acquiring a surface image of the metal piece to be evaluated, and performing image processing on the surface image to obtain a gray level image.
The CCD camera is used for collecting RGB images on the surface of the aluminum alloy rod through overlooking view angles, so that the clustering radius can be accurately determined conveniently, the required ambient illumination is uniform in the image collecting process, and other influencing factors such as foreign matter shielding are avoided. FIG. 2 shows an image of a portion of an RGB image acquired of the surface of an aluminum alloy rod having a pinhole defect. And carrying out corresponding image processing, such as denoising, graying and other conventional processing, on the acquired RGB image, thereby obtaining a corresponding gray image.
Step S2: and determining each target defect pixel point according to the gray value of each pixel point in the gray image, and carrying out aggregation grouping on each target defect pixel point according to the position of each target defect pixel point to obtain each aggregation block and the central defect point of each aggregation block.
When the DBSCAN density clustering algorithm is adopted to carry out region segmentation on the gray level image of the aluminum alloy rod, different segmentation regions are mainly obtained according to gray level changes in the image. Because the air hole defects formed on the surface of the aluminum alloy rod are not singly distributed, are usually continuously and densely distributed, and different defect areas can be connected together or are close to each other, the selection of a proper clustering radius is particularly critical when the pixels are clustered through a DBSCAN clustering algorithm. If the selected cluster radius is too large, the pixel points in different areas are gathered into one type, so that two or more small air holes are divided into one large air hole, and the error is larger when the quality of the aluminum alloy rod is evaluated; if the selected cluster radius is too small, pixel points originally belonging to the same area can be separated, so that the defect area of the image is increased. Therefore, when the gray level image of the aluminum alloy rod is segmented by adopting a DBSCAN density clustering algorithm, obtaining a proper clustering radius is important.
In order to facilitate the subsequent acquisition of a suitable cluster radius, determining each target defect pixel point according to the gray value of each pixel point in the gray image, the specific implementation steps include:
step S21: and determining a gray level histogram according to the gray level value of each pixel point in the gray level image, and determining a gray level curve according to the gray level histogram.
Step S22: and determining target peaks in the peaks on the gray scale curves, and determining the maximum gray scale value corresponding to the target peaks.
Step S23: and determining each target defect pixel point in the gray level image according to the maximum gray level value corresponding to the target peak, wherein the gray level value of the target defect pixel point is equal to the maximum gray level value.
Specifically, according to the gray values of each pixel point in the gray map obtained in step S1, statistics is performed on the occurrence frequency of each gray value, so that a gray histogram can be obtained. Since the specific implementation process of acquiring the gray histogram belongs to the prior art, the description is omitted here. After the gray level histogram is obtained, each gray level value in the gray level histogram is taken as an abscissa, and a frequency value corresponding to each gray level value is taken as an ordinate, and a curve is fitted, so that a gray level curve can be obtained.
Because the gray value of the air hole defect on the surface of the aluminum alloy rod is compared with the gray value of the normal areaSmaller, so that two peaks appear in the obtained gray scale curve, the small peak represents the gray scale value of the pixel point of the suspected air hole defect area, the large peak represents the gray scale value of the pixel point of the normal area, the small peak is selected as the target peak in each peak on the gray scale curve, and the gray scale value corresponding to the maximum frequency in the target peak is determined
Gray value corresponding to maximum frequency in the target peak
The maximum gray value corresponding to the target peak is obtained.
And determining each pixel point with the gray value being the maximum gray value in the gray image according to the maximum gray value corresponding to the target peak, wherein the pixel points represent the pixel points in the suspected air hole defect area but are not complete air hole defect areas, and the pixel points are called target defect pixel points and are also called mark pixel points. Because the gray values of the pixels of the suspected air hole defect area are similar, the obtained target defect pixels are scattered in the image.
After each target defective pixel point in the gray-scale image is obtained through the above steps S21 to S23, the target defective pixel points are subjected to density clustering, that is, aggregation grouping, so that each aggregation block can be obtained. When density clustering is performed on the target defect pixel points, the used clustering radius is usually set smaller, the clustering radius is set to be 2 in the embodiment, and the clustering radius can be adaptively modified according to requirements by a person skilled in the art. It should be noted that, when the density clustering is performed on the target defective pixels, since the target defective pixels are known pixels, after determining a suitable cluster radius, the aggregate of different densities can be obtained.
After each aggregation block is obtained, a center defect point corresponding to each aggregation block is determined. In this embodiment, the central defect point of each aggregation block is a clustering central point corresponding to each aggregation block obtained by aggregating and grouping each target defect pixel point, that is, when the target defect pixel points are clustered in density, each clustering center corresponding to each aggregation block is used as the central defect point of the corresponding aggregation block.
Step S3: and determining the gradient mean value corresponding to each pixel point on the central defect point connecting line of every two aggregation blocks according to the gray values of each pixel point and the surrounding neighborhood pixel points on the central defect point connecting line of every two aggregation blocks.
Since each of the above-mentioned clusters is obtained by density-clustering target defective pixels of the same gray scale, each of the clusters thus obtained is only a part of the pixels in the suspected pore defect region, and the density clusters are clustered according to the aggregability of the pixels, and the discrete points belonging to the region cannot be clustered into one type. Therefore, a complete suspected pore defect region can be obtained by using a DBSCAN clustering method, and the key of using the DBSCAN clustering algorithm is to select a proper clustering radius.
When determining the clustering radius of the DBSCAN clustering algorithm, the fact that the distance between the suspected pore defect areas on the aluminum alloy rod is smaller is considered, so that when the clustering radius is calculated, the distance between different suspected pore defect areas needs to be obtained according to edge pixel points of the suspected pore defect areas, and the smaller the distance is, the smaller the clustering radius is required when the pixel points in the image are clustered, and therefore the pixel points originally belonging to the two suspected pore defect areas cannot be clustered into one type. Therefore, when the DBSCAN clustering algorithm is adopted to cluster images later, the clustering radius needs to be adjusted according to the distance between the suspected air hole defect areas which are close to each other.
Based on the above analysis, in order to measure the distance between different suspected air hole defect areas, the present embodiment proposes a concept of a relevance index, where the relevance index characterizes the distance between the suspected air hole defect areas where any two aggregation blocks are located. When the distance between the suspected air hole defect areas where any two aggregation blocks are located is smaller, the suspected air hole defect areas where the two aggregation blocks are located are in an adhesion state, and at the moment, the relevance index of the two aggregation blocks is larger; when the distance between the suspected air hole defect areas where any two aggregation blocks are located is larger, the suspected air hole defect areas where the two aggregation blocks are located are independent of each other, and at the moment, the relevance index of the two aggregation blocks is smaller.
In order to facilitate the subsequent determination of the relevance index between any two aggregation blocks, according to the gray values of each pixel point on the central defect point connecting line of each two aggregation blocks and the neighboring pixel points thereof, the gradient mean value corresponding to each pixel point on the central defect point connecting line of each two aggregation blocks is determined, and the specific implementation steps comprise:
step S31: and determining the gray scale gradient of each pixel point and the neighborhood pixel points around the pixel points on the connecting line of the central defect points of each two aggregation blocks according to the gray scale values of each pixel point and the neighborhood pixel points around the pixel points on the connecting line of the central defect points of each two aggregation blocks.
Step S32: and constructing a sliding window area of each pixel point on the connecting line of the central defect points of every two aggregation blocks, and determining the average value of gray gradient of each pixel point in the sliding window area as the average value of gradient corresponding to the pixel point on the connecting line corresponding to the sliding window area.
Considering that the gradient value of the pixel point at the edge of the air hole defect area is larger, if the distance between the two aggregation blocks is smaller, the gradient change of the pixel point at the edge is more complex and is not a single change. Since the gradation from one cluster block to another cluster block may change a plurality of times, the correlation of any two cluster blocks can be obtained from the gradient change of the pixel points between the two cluster blocks.
To obtain the association of any two aggregate blocks, first, the association is obtained by
The operator obtains the gradient magnitude of each pixel point in the gray image, namely the gray gradient. For any two aggregation blocks, the center defect points of the two aggregation blocks are connectedAnd a sliding window 3*3 is arranged at the rear, the center of the sliding window is a pixel point on a connecting line between the center defect points of the two aggregation blocks, the sliding window slides one pixel point at a time along the connecting line between the center defect points of the two aggregation blocks, and the sliding direction of the sliding window is marked. In the sliding process of the sliding windows, calculating the average gray gradient of all pixel points in each sliding window, and taking the average gray gradient as the gradient mean value corresponding to the pixel points on the connecting line corresponding to the center of the sliding window. In this way, a gradient average value corresponding to each pixel point on the line between the center defect points of the two aggregation blocks can be obtained.
Step S4: and determining the relevance index of each two aggregation blocks according to the gradient mean value corresponding to each pixel point on the central defect point connecting line of each two aggregation blocks and the distance value between the central defect points of each two aggregation blocks.
For any two aggregation blocks, calculating a distance value between central defect points of the aggregation blocks, wherein the distance value refers to Euclidean distance, and then determining a relevance index of each two aggregation blocks based on the step S3, wherein the specific steps comprise:
step S41: and sequencing the gradient mean values corresponding to the pixel points on the central defect point connecting lines of every two aggregation blocks according to the direction from one end to the other end of the central defect point connecting lines of every two aggregation blocks, so as to obtain a gradient mean value sequence of every two aggregation blocks.
Step S42: determining a difference value between any two adjacent gradient mean values in the gradient mean value sequence of every two aggregation blocks, determining a target gradient mean value according to two gradient mean values corresponding to the difference value between the largest two adjacent gradient mean values, and determining a relevance index of every two aggregation blocks according to the difference value between the largest two adjacent gradient mean values corresponding to every two aggregation blocks, the target gradient mean value and the distance value between the center defect points of every two aggregation blocks, wherein the relevance index and the difference value between the corresponding largest two adjacent gradient mean values, the target gradient mean value and the distance value are all in a negative correlation relationship.
Specifically, for any twoAnd the aggregation blocks sort the gradient mean values obtained by each sliding window according to the sliding direction of the sliding windows, so that a gradient mean value sequence can be obtained. Determining any one gradient mean value in the gradient mean value sequence
Mean value of gradient before it
And determining the maximum value of all the differences corresponding to any two aggregation blocks, and taking the next gradient mean value corresponding to the maximum difference as a target gradient mean value. Based on the difference value between the largest two adjacent gradient mean values corresponding to any two aggregation blocks, the target gradient mean value and the distance value, the relevance index of the two aggregation blocks can be determined, and the corresponding calculation formula is as follows:
wherein ,
for the relevance index of every two aggregation blocks,
for the difference between the largest two adjacent gradient averages corresponding to each two aggregate blocks,
for each two corresponding target gradient averages of the aggregate block,
for the distance value between the center defect points of every two aggregation blocks,expto be with natural constanteAn exponential function of the base, the function of which is to
Reverse normalized to the range of 0-1.
The setting logic of the relevance index is as follows: for two suspected air hole defect areas with closer distances, the requirement on the clustering radius is more strict in the segmentation process, and if the clustering radius is larger, the two suspected air hole defect areas are segmented into one block. Therefore, according to the distance value between the two aggregation blocks as a judging condition, the distance value represents the Euclidean distance between the center defect points of the two aggregation blocks, and if the distance value between the two aggregation blocks is smaller, the relevance between the two aggregation blocks is larger, and the cluster radius obtained by the aggregation blocks with larger relevance is smaller. Meanwhile, when the two suspected air hole defect areas are close to each other, the two suspected air hole defect areas may overlap, so that the edge gradient of the suspected air hole defect areas is reduced and the gradient difference is reduced. Therefore, by considering the difference between the largest two adjacent gradient averages corresponding to any two aggregation blocks, the target gradient average value and the distance value can represent the relevance of the two aggregation blocks, when the difference between the largest two adjacent gradient averages, the target gradient average value and the distance value are smaller, the relevance of the two aggregation blocks is larger, and at the moment, the corresponding clustering radius is smaller when an image is segmented.
It should be noted that, in the above-mentioned calculation formula of the relevance index, the present embodiment only considers the difference between the largest two adjacent gradient averages, the target gradient average and the logical relationship between the distance value and the relevance index, and does not consider the dimension problem between each other. And evaluating the characteristics between the two suspected air hole defect areas where the two aggregation blocks are positioned by using the relevance index between the two aggregation blocks, wherein if the relevance index is larger, the closer the clusters between the two suspected air hole defect areas are, the smaller the corresponding cluster radius is, and the defect areas of the air Kong Yishi can be separated.
Step S5: and determining the aggregation index of each two aggregation blocks according to the gradient mean value corresponding to each pixel point on the central defect point connecting line of each two aggregation blocks and the association index of each two aggregation blocks.
Fig. 3 shows the positional relationship between the suspected air hole defect regions where different aggregation blocks are located, in fig. 3, the suspected air hole defect region where the aggregation block a is located and the suspected air hole defect region where the aggregation block B is located are in an adjacent state, the distance between the suspected air hole defect region where the aggregation block a is located and the suspected air hole defect region where the aggregation block C is located is large, and there is an overlapping portion between the suspected air hole defect region where the aggregation block C is located and the suspected air hole defect region where the aggregation block D is located.
When the edge position relations of the different suspected air hole defect areas are different, the gray level change of the edge pixel points is also different, if the position relation between the suspected air hole defect area where the aggregation block A is positioned and the suspected air hole defect area where the aggregation block B is positioned is the same, the gradient of the edges of the two areas is larger, and only one edge mutation point appears; if the position relationship between the suspected pore defect region where the aggregation block C is located and the suspected pore defect region where the aggregation block D is located is the same, the edges of the two regions are weakened because the superposition of different regions occurs, and the edge gradient is small; if the position relationship between the suspected air hole defect region where the aggregation block B is located and the suspected air hole defect region where the aggregation block E is located is the same, the error division is avoided because the distance between the two regions is far, so that the consideration is not needed. Therefore, in determining the cluster radius, not only the distance between the two suspected air hole defect areas, but also the edge position relationship between the suspected air hole defect areas needs to be considered, so as to divide the two suspected air hole defect areas where the aggregation blocks a and B are located and the two suspected air hole defect areas where the aggregation blocks C and D are located in fig. 3.
The correlation between the two suspected air hole defect areas can be represented by the correlation index between the aggregation blocks, but the change between the two area edges cannot be represented, and the position relationship of the two area edges is related to the size of the clustering radius, so that the change relationship of the aggregation block edges is considered when the clustering is carried out. In order to facilitate the subsequent determination of a suitable aggregation radius, the present embodiment proposes a concept of an aggregation index, where the aggregation index not only considers the relevance between aggregation blocks, but also considers the positional relationship of edge pixels of the suspected pore defect area where any two aggregation blocks are located, and when determining such positional relationship, the present embodiment does not actually obtain edges of different areas, but describes states of edges of different suspected pore defect areas according to gray scales of the pixels between the two aggregation blocks, and the specific implementation process includes:
Step S51: for the gradient mean value sequence of every two aggregation blocks, determining the maximum gradient mean value in the gradient mean value sequence, calculating the mean value of the difference values between the maximum gradient mean value and each gradient mean value in front of the gradient mean value sequence, thereby obtaining a first gradient difference index, and calculating the mean value of the difference values between the maximum gradient mean value and each gradient mean value behind the gradient mean value, thereby obtaining a second gradient difference index.
Step S52: according to the relevance index, the first gradient difference index and the second gradient difference index of the gradient mean value sequence of every two aggregation blocks, determining the aggregation index of every two aggregation blocks, wherein the absolute value of the difference value and the relevance index of the aggregation index and the corresponding first gradient difference index and second gradient difference index are positive correlation relations.
Specifically, for any two aggregation blocks, determining the maximum gradient mean value in a gradient mean value sequence corresponding to the two aggregation blocks, marking each gradient mean value positioned in front of the maximum gradient mean value in the sequence of the maximum gradient mean value as a front gradient mean value, marking each gradient mean value positioned behind the maximum gradient mean value in the sequence of the maximum gradient mean value as a rear gradient mean value, calculating the average value of the difference between the maximum gradient mean value and each front gradient mean value, calculating the average value of the difference between the maximum gradient mean value and each rear gradient mean value, taking the average value of the previous difference value as a first gradient difference index, and taking the average value of the next difference value as a second gradient difference index. Based on the relevance index, the first gradient difference index and the second gradient difference index of the gradient mean value sequence of the two aggregation blocks, determining the aggregation index of the two aggregation blocks, wherein the corresponding calculation formula is as follows:
wherein ,
for the aggregate index of every two aggregate blocks,
is the maximum gradient mean value in the gradient mean value sequence,
is the maximum gradient mean value in the gradient mean value sequence
Is the first one of the followingiThe average value of the individual gradients,
is the maximum gradient mean value in the gradient mean value sequence
Is the following ofrThe average value of the gradients, n and m are respectively the maximum average value of the gradients in the gradient average value sequence
The total number of preceding and following gradient averages,
for the first gradient difference index corresponding to every two aggregation blocks,
for the second gradient difference index corresponding to every two aggregation blocks, e is a natural constant,expto be with natural constanteAn exponential function of the base for
Inversely normalized to the range of 0-1, K is the relevance index of every two aggregation blocks.
The above-mentioned setting logic of the aggregation index is: the gradient mean value sequence characterizes the gray level change condition of the pixel points between the two aggregation blocks, and the pixel point corresponding to the maximum gradient mean value is the edge pixel point of the two suspected air hole defect areas where the two aggregation blocks are located, so that the position relationship of the two suspected air hole defect areas where the two aggregation blocks are located can be represented according to the integral gray level change of the pixel points at the left side and the right side of the pixel point corresponding to the maximum gradient mean value. If the two suspected air hole defect areas are the position relations of the two suspected air hole defect areas where the aggregation blocks A and B are located in the diagram or the position relations of the two suspected air hole defect areas where the aggregation blocks C and D are located, the value of the aggregation index W is larger, and the gradient difference of the left side and the right side of the pixel point corresponding to the maximum gradient mean value is not large, so that the superposition of the pixel point with the maximum gradient mean value is counteracted. If the two suspected air hole defect areas are the position relations of the two suspected air hole defect areas where the aggregation blocks B and E are located, the value of the aggregation index W is smaller, and the gradient average value on one side of the maximum gradient average value in the gradient average value sequence is definitely larger than the gradient average value on the other side because only the edge pixel point corresponding to the maximum value of one gradient average value is selected and the edge of the other suspected air hole defect area is still arranged on the connecting line of the two aggregation blocks. Therefore, when the optimal clustering radius is obtained according to the aggregation index, the appropriate clustering radius can be obtained according to the position relation of the suspected air hole defect areas where the two aggregation blocks are located, the suspected air hole defect areas cannot be misclassified, and finally the accuracy of quality assessment is guaranteed.
According to the embodiment, the position relation of the two suspected air hole defect areas is represented by adopting the aggregation indexes of the two aggregation blocks, when the distance between the two suspected air hole defect areas is relatively short and the gray level difference of the edge pixel points of the two suspected air hole defect areas is relatively small, the fact that the two suspected air hole defect areas are possibly overlapped is indicated, and the corresponding aggregation indexes are relatively large; when the distance between the two suspected air hole defect areas is far and the gray level difference of the edge pixel points of the two suspected air hole defect areas is large, the two suspected air hole defect areas are not overlapped, and the corresponding aggregation index is small. Therefore, the evaluation of the position relationship of the two suspected air hole defect areas is not only described by the distance, but also is evaluated according to the gray level change of the pixel point between the two, so that the position relationship of different suspected air hole defect areas can be more accurately determined, and the subsequent more accurate determination of the aggregation radius is facilitated.
Step S6: and determining each candidate clustering radius according to the aggregation index of each two aggregation blocks, the gray value of each pixel point on the central defect point connecting line of each two aggregation blocks and the gradient mean value corresponding to each pixel point.
Through the step S5, the aggregate index of any two aggregate blocks can be obtained, and when the aggregate index is larger, the closer the distance between the two corresponding suspected air hole defect areas is, the more likely the edges of the two suspected air hole defect areas overlap, and the smaller the required clustering radius is, so that the pixel points on the two areas can be prevented from being misclassified. In addition, considering that the edge gray value of the air hole defect area is larger than the center gray value, when the difference between the edge gray value and the center gray value of the area is smaller, the weakening degree of the edge of the area is larger, so that smaller clustering radius is needed to be selected when the clustering radius is selected, the situation that pixels of different areas are wrongly clustered into one type can be avoided, and the edge of the obtained segmented area can be clearer. Therefore, based on the aggregation index of every two aggregation blocks, the gray value of the center defect point of every two aggregation blocks, the gray value of each pixel point on the connecting line of the center defect points of every two aggregation blocks and the gradient mean value corresponding to each pixel point, the method determines each candidate clustering radius, and specifically comprises the following steps:
step S61: and determining the maximum value in the gradient mean value corresponding to each pixel point on the central defect point connecting line of every two aggregation blocks according to the gray value of each pixel point on the central defect point connecting line of every two aggregation blocks and the gradient mean value corresponding to each pixel point, and determining the gray value of the pixel point on the connecting line corresponding to the maximum value in the gradient mean value as the edge gray value of every two aggregation blocks.
Step S62: and determining candidate cluster radiuses corresponding to every two aggregation blocks according to the gray values of the center defect points of every two aggregation blocks, the edge gray values of every two aggregation blocks and the aggregation index, wherein the difference value between the candidate cluster radiuses and the gray values of the corresponding center defect points and the difference value between the candidate cluster radiuses and the edge gray values are positive correlation, and the candidate cluster radiuses and the corresponding aggregation index are negative correlation.
In this embodiment, for any two aggregation blocks, the calculation formula corresponding to the determined candidate cluster radius is:
wherein ,
for each two cluster block corresponding candidate cluster radii,
for the edge gray value of every two aggregation blocks,
for the gray value of the center defect point of every two aggregation blocks,
for the aggregate index of every two aggregate blocks, e is a natural constant,
for inversely normalizing the aggregation index W to be in the range of 0-1.
The setting logic of the candidate cluster radius is as follows: due to aggregationThe index W represents the position relation of two suspected air hole defect areas, and when the aggregation index W is larger, the closer the distance between the two suspected air hole defect areas is and the more possible overlapping is indicated, the smaller the required clustering radius is, and the wrong division of pixel points on the two areas can be avoided. Meanwhile, when the difference of gray values between the edge and the center of the region is smaller, that is 
The smaller the area, the greater the weakening degree of the edge of the area, namely, the smaller the difference is, the closer the gray values of the center pixel point and the edge pixel point are, the smaller the difference between the edge area and the center pixel point is, and the greater the weakening degree of the edge is, the smaller the clustering radius is needed, so that the pixels on the two areas can be correctly distinguished.
Step S7: and determining an optimal density clustering radius according to each candidate clustering radius, and performing density clustering on each pixel point in the gray level image according to the optimal density clustering radius, so as to obtain a defect area in the gray level image, and further determining a quality evaluation result of the metal piece.
After each candidate cluster radius is obtained through the above step S6, the smallest candidate cluster radius among the candidate cluster radii is determined as the optimal density cluster radius, i.e., the smallest cluster radius among the candidate cluster radii is selected and determined as the optimal density cluster radius. According to the optimal density clustering radius, and the proper clustering pixel number is set, an implementer can determine the proper clustering pixel number according to experiments or experience and combining practical situations, the embodiment sets the clustering pixel number to be 4, and meanwhile, based on the gray value and the position of each pixel in the gray image, the gray image is clustered through a DBSCAN clustering algorithm, and the pixels belonging to the same area are classified into one type, so that the air hole defect area in the gray image is determined. Because the specific implementation process of the DBSCAN density clustering algorithm belongs to the prior art, the focus of the embodiment is on accurately determining the clustering radius of the DBSCAN density clustering algorithm, and the specific implementation steps of the DBSCAN density clustering algorithm are not repeated here.
After the air hole defect area in the gray level image is obtained through the DBSCAN clustering algorithm, the air hole defect area is marked so as to facilitate manual verification. And determining the production quality condition of the aluminum alloy rod according to the distribution condition of each air hole defect area. The quality of the aluminum alloy rod is evaluated according to the defect areas of each air hole, and the quality can be determined according to the requirements of each manufacturer. For example, when the number of the air hole defect areas on the surface of the aluminum alloy rod exceeds a certain number or the total area of the air hole defect areas exceeds a certain area value, the aluminum alloy rod is judged to be unqualified in production quality.
According to the method, the surface image of the aluminum alloy rod is obtained, and the abnormal pixel point, namely the target defect pixel point, is obtained through the gray level change of the pixel point in the gray level image corresponding to the surface image. Each aggregation block is determined according to the target defect pixel point, the relevance index between the different aggregation blocks is determined by considering the distance between the different aggregation blocks, then the aggregation index between the different aggregation blocks is determined by combining the edge gray scale change condition of the suspected air hole defect areas corresponding to the different aggregation blocks, and the aggregation index can accurately represent the position relation of the suspected air hole defect areas corresponding to the different aggregation blocks. According to the aggregation indexes among different aggregation blocks, the gray scale relation between the center and the edge positions of the suspected air hole defect area corresponding to the aggregation blocks is combined, each candidate cluster radius is determined, and finally the optimal density cluster radius is screened out. Because the optimal density clustering radius is matched with the gray level image corresponding to the surface image of the aluminum alloy rod, when the DBSCAN density clustering algorithm is adopted to perform density clustering on each pixel point in the gray level image, the air hole defect area in the gray level image can be accurately divided, and therefore the accuracy of the production quality assessment of the aluminum alloy rod is finally improved.
It should be noted that: the above embodiments are only for illustrating the technical solution of the present application, and are not limiting; 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 scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present application, and are intended to be included in the scope of the present application.
Claims (7)
1. An aluminum alloy rod production quality evaluation method based on image processing is characterized by comprising the following steps:
acquiring a surface image of a metal piece to be evaluated, and performing image processing on the surface image to obtain a gray level image;
determining each target defect pixel point according to the gray value of each pixel point in the gray image, and carrying out aggregation grouping on each target defect pixel point according to the position of each target defect pixel point to obtain each aggregation block and the central defect point of each aggregation block;
according to the gray values of each pixel point and the neighborhood pixel points around the pixel points on the central defect point connecting line of every two aggregation blocks, determining the gradient mean value corresponding to each pixel point on the central defect point connecting line of every two aggregation blocks;
Determining the relevance index of each two aggregation blocks according to the gradient mean value corresponding to each pixel point on the central defect point connecting line of each two aggregation blocks and the distance value between the central defect points of each two aggregation blocks;
determining the aggregation index of each two aggregation blocks according to the gradient mean value corresponding to each pixel point on the central defect point connecting line of each two aggregation blocks and the association index of each two aggregation blocks;
determining each candidate clustering radius according to the aggregation index of each two aggregation blocks, the gray value of each pixel point on the central defect point connecting line of each two aggregation blocks and the gradient mean value corresponding to each pixel point;
determining an optimal density clustering radius according to each candidate clustering radius, and performing density clustering on each pixel point in the gray level image according to the optimal density clustering radius, so as to obtain a defect area in the gray level image, and further determining a quality evaluation result of the metal piece;
determining an association index for each two aggregation blocks, comprising:
sequencing the gradient mean values corresponding to the pixel points on the central defect point connecting lines of every two aggregation blocks according to the direction from one end to the other end of the central defect point connecting lines of every two aggregation blocks, so as to obtain a gradient mean value sequence of every two aggregation blocks;
Determining a difference value between any two adjacent gradient mean values in a gradient mean value sequence of every two aggregation blocks, determining a target gradient mean value according to two gradient mean values corresponding to the difference value between the largest two adjacent gradient mean values, and determining a relevance index of every two aggregation blocks according to the difference value between the largest two adjacent gradient mean values corresponding to every two aggregation blocks, the target gradient mean value and a distance value between center defect points of every two aggregation blocks, wherein the relevance index and the difference value between the corresponding largest two adjacent gradient mean values, the target gradient mean value and the distance value are all in a negative correlation relationship;
determining an aggregate index for each two aggregate blocks, comprising:
for the gradient mean value sequence of every two aggregation blocks, determining the maximum gradient mean value in the gradient mean value sequence, calculating the mean value of the difference values between the maximum gradient mean value and each gradient mean value in front of the gradient mean value sequence, thereby obtaining a first gradient difference index, and calculating the mean value of the difference values between the maximum gradient mean value and each gradient mean value behind the gradient mean value, thereby obtaining a second gradient difference index;
determining the aggregation index of each two aggregation blocks according to the corresponding correlation index, the first gradient difference index and the second gradient difference index of each two aggregation blocks, wherein the absolute difference value and the correlation index of the aggregation index and the corresponding first gradient difference index and second gradient difference index are positive correlation;
Determining respective candidate cluster radii, comprising:
according to the gray values of all the pixel points on the connecting line of the central defect points of every two aggregation blocks and the gradient average values corresponding to all the pixel points, determining the maximum value in the gradient average values corresponding to all the pixel points on the connecting line of the central defect points of every two aggregation blocks, and determining the gray value of the pixel point on the connecting line corresponding to the maximum value in the gradient average values as the edge gray value of every two aggregation blocks;
and determining candidate cluster radiuses corresponding to every two aggregation blocks according to the gray values of the center defect points of every two aggregation blocks, the edge gray values of every two aggregation blocks and the aggregation index, wherein the difference values of the candidate cluster radiuses and the gray values of the corresponding center defect points and the edge gray values are positive correlation relations, and the candidate cluster radiuses and the corresponding aggregation index are negative correlation relations.
2. The image processing-based aluminum alloy rod production quality evaluation method according to claim 1, wherein determining each target defective pixel comprises:
determining a gray level histogram according to the gray level value of each pixel point in the gray level image, and determining a gray level curve according to the gray level histogram;
determining target peaks in the peaks on the gray scale curves, and determining maximum gray scale values corresponding to the target peaks;
And determining each target defect pixel point in the gray level image according to the maximum gray level value corresponding to the target peak, wherein the gray level value of the target defect pixel point is equal to the maximum gray level value.
3. The image processing-based aluminum alloy rod production quality evaluation method according to claim 1, wherein determining a gradient mean value corresponding to each pixel point on a central defect point connecting line of each two aggregation blocks comprises:
according to the gray values of each pixel point and the neighborhood pixel points around the pixel points on the central defect point connecting line of every two aggregation blocks, determining the gray gradient of each pixel point and the neighborhood pixel points around the pixel points on the central defect point connecting line of every two aggregation blocks;
and constructing a sliding window area of each pixel point on the connecting line of the central defect points of every two aggregation blocks, and determining the average value of gray gradient of each pixel point in the sliding window area as the average value of gradient corresponding to the pixel point on the connecting line corresponding to the sliding window area.
4. The image processing-based aluminum alloy rod production quality evaluation method according to claim 1, wherein the calculation formula corresponding to the correlation index of each two aggregation blocks is determined as follows:
wherein ,
for the relevance index of every two aggregation blocks, < +. >
For the difference between the largest two adjacent gradient means corresponding to every two aggregate blocks,/>
For the target gradient mean value corresponding to every two aggregate blocks, < >>
For the distance value between the center defect points of every two aggregation blocks,expto be with natural constanteIs an exponential function of the base.
5. The image processing-based aluminum alloy rod production quality evaluation method according to claim 1, wherein the calculation formula corresponding to the aggregation index determining each two aggregation blocks is:
wherein ,
for the aggregate index of every two aggregate blocks, -/->
For the first gradient difference index corresponding to every two aggregate blocks,/>
For the second gradient difference index corresponding to every two aggregation blocks, e is a natural constant,expto be with natural constanteK is the relevance index of every two aggregation blocks as an exponential function of the base.
6. The image processing-based aluminum alloy rod production quality evaluation method according to claim 1, wherein determining an optimal density cluster radius comprises:
and determining the smallest candidate cluster radius in the candidate cluster radii as the optimal density cluster radius.
7. The image processing-based aluminum alloy rod production quality evaluation method according to claim 1, wherein the center defect point of each aggregation block is a cluster center point corresponding to when each target defect pixel point is aggregated and grouped to obtain each aggregation block.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310005137.5A CN115690108B (en) | 2023-01-04 | 2023-01-04 | Aluminum alloy rod production quality assessment method based on image processing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310005137.5A CN115690108B (en) | 2023-01-04 | 2023-01-04 | Aluminum alloy rod production quality assessment method based on image processing |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115690108A CN115690108A (en) | 2023-02-03 |
| CN115690108B true CN115690108B (en) | 2023-04-28 |
Family
ID=85056997
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310005137.5A Active CN115690108B (en) | 2023-01-04 | 2023-01-04 | Aluminum alloy rod production quality assessment method based on image processing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115690108B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116091504B8 (en) * | 2023-04-11 | 2023-09-15 | 重庆大学 | Connecting pipe connector quality detection method based on image processing |
| CN116152247B (en) * | 2023-04-20 | 2023-08-25 | 江苏三纳科技材料有限公司 | Quality evaluation method of polyurethane coating film |
| CN116596936B (en) * | 2023-07-18 | 2023-09-12 | 深圳市魔方卫星科技有限公司 | Solar wing sailboard quality detection method based on image data |
| CN116912241B (en) * | 2023-09-12 | 2023-12-12 | 深圳市艾为创科技有限公司 | CNC machine adjustment optimization method and system based on machine learning |
| CN116912256B (en) * | 2023-09-14 | 2023-11-28 | 山东大昌纸制品有限公司 | Corrugated paper rib defect degree assessment method based on image processing |
| CN116993726B (en) * | 2023-09-26 | 2023-12-19 | 山东克莱蒙特新材料科技有限公司 | Mineral casting detection method and system |
| CN117409373B (en) * | 2023-12-14 | 2024-02-09 | 艾信智慧医疗科技发展（苏州）有限公司 | Medical logistics box full-period intelligent monitoring system |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114862862A (en) * | 2022-07-11 | 2022-08-05 | 江苏大田阀门制造有限公司 | Pump body cold shut defect identification method and system based on image processing |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101330098B1 (en) * | 2012-05-10 | 2013-11-18 | 동우 화인켐 주식회사 | Method for discriminating defect of optical films |
| CN113706537B (en) * | 2021-10-28 | 2022-02-25 | 江苏启力锻压机床有限公司 | Irregular workpiece detection method based on machine vision |
| CN115115613B (en) * | 2022-07-26 | 2023-04-07 | 南通博莹机械铸造有限公司 | Paint spraying defect detection method and system based on machine vision |
| CN114943739B (en) * | 2022-07-26 | 2022-10-21 | 山东三微新材料有限公司 | Aluminum pipe quality detection method |
| CN115115621B (en) * | 2022-08-24 | 2022-11-11 | 聊城市泓润能源科技有限公司 | Lubricating oil pollution degree detection method based on image processing |
| CN115311292B (en) * | 2022-10-12 | 2023-01-17 | 南通创铭伊诺机械有限公司 | Strip steel surface defect detection method and system based on image processing |
-
2023
- 2023-01-04 CN CN202310005137.5A patent/CN115690108B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114862862A (en) * | 2022-07-11 | 2022-08-05 | 江苏大田阀门制造有限公司 | Pump body cold shut defect identification method and system based on image processing |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115690108A (en) | 2023-02-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115690108B (en) | Aluminum alloy rod production quality assessment method based on image processing | |
| CN115601364B (en) | Golden finger circuit board detection method based on image analysis | |
| CN115082418B (en) | Precise identification method for automobile parts | |
| CN116109644B (en) | Surface defect detection method for copper-aluminum transfer bar | |
| CN115345885A (en) | Method for detecting appearance quality of metal fitness equipment | |
| CN115439474B (en) | Rapid positioning method for power equipment fault | |
| CN115511889B (en) | Method for detecting welding defects on surface of solar cell panel bracket | |
| CN116309537B (en) | Defect detection method for oil stain on surface of tab die | |
| CN115829883A (en) | Surface image denoising method for dissimilar metal structural member | |
| CN115294159B (en) | Method for dividing corroded area of metal fastener | |
| CN115861307B (en) | Fascia gun power supply driving plate welding fault detection method based on artificial intelligence | |
| CN113554629A (en) | Strip steel red rust defect detection method based on artificial intelligence | |
| CN114926463A (en) | Production quality detection method suitable for chip circuit board | |
| CN115719332A (en) | Welding quality detection method | |
| CN114782329A (en) | Bearing defect damage degree evaluation method and system based on image processing | |
| CN115131359B (en) | Method for detecting pitting defects on surface of metal workpiece | |
| CN116523923B (en) | Battery case defect identification method | |
| CN115311484B (en) | Optimized clustering segmentation method for weld data | |
| CN116152242B (en) | Visual detection system of natural leather defect for basketball | |
| CN115984272B (en) | Semitrailer axle defect identification method based on computer vision | |
| CN114820625A (en) | Automobile top block defect detection method | |
| CN115908362A (en) | Method for detecting wear resistance of skateboard wheel | |
| CN117094975A (en) | Method and device for detecting surface defects of steel and electronic equipment | |
| CN115619708A (en) | Method for judging fault based on image recognition of oil quality change of main shaft bearing lubricating oil | |
| CN116977230B (en) | Scanning electron microscope image optimization enhancement method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |