CN115965624A - Detection method for anti-wear hydraulic oil pollution particles - Google Patents
Detection method for anti-wear hydraulic oil pollution particles Download PDFInfo
- Publication number
- CN115965624A CN115965624A CN202310252093.6A CN202310252093A CN115965624A CN 115965624 A CN115965624 A CN 115965624A CN 202310252093 A CN202310252093 A CN 202310252093A CN 115965624 A CN115965624 A CN 115965624A
- Authority
- CN
- China
- Prior art keywords
- pixel
- flow direction
- hydraulic oil
- block
- point pair
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 239000010720 hydraulic oil Substances 0.000 title claims abstract description 89
- 239000002245 particle Substances 0.000 title claims abstract description 73
- 238000001514 detection method Methods 0.000 title abstract description 15
- 238000000034 method Methods 0.000 claims abstract description 37
- 238000012545 processing Methods 0.000 claims abstract description 19
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 claims description 35
- 229910052742 iron Inorganic materials 0.000 claims description 17
- 238000010606 normalization Methods 0.000 claims description 12
- 238000001228 spectrum Methods 0.000 claims description 11
- 238000013507 mapping Methods 0.000 claims description 9
- 238000013528 artificial neural network Methods 0.000 claims description 7
- 238000011109 contamination Methods 0.000 claims description 2
- 239000006249 magnetic particle Substances 0.000 description 21
- 238000004364 calculation method Methods 0.000 description 9
- 230000005291 magnetic effect Effects 0.000 description 9
- 230000000694 effects Effects 0.000 description 7
- 239000008187 granular material Substances 0.000 description 4
- 238000011156 evaluation Methods 0.000 description 3
- 238000003708 edge detection Methods 0.000 description 2
- 230000002708 enhancing effect Effects 0.000 description 2
- 230000005294 ferromagnetic effect Effects 0.000 description 2
- 229910052751 metal Inorganic materials 0.000 description 2
- 239000002184 metal Substances 0.000 description 2
- 239000004576 sand Substances 0.000 description 2
- 241001270131 Agaricus moelleri Species 0.000 description 1
- 229910000831 Steel Inorganic materials 0.000 description 1
- 238000005299 abrasion Methods 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
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 229960001948 caffeine Drugs 0.000 description 1
- 239000000919 ceramic Substances 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000005389 magnetism Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000002923 metal particle Substances 0.000 description 1
- 239000003921 oil Substances 0.000 description 1
- 230000003647 oxidation Effects 0.000 description 1
- 238000007254 oxidation reaction Methods 0.000 description 1
- 239000013618 particulate matter Substances 0.000 description 1
- 239000010959 steel Substances 0.000 description 1
- 230000002194 synthesizing effect Effects 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- RYYVLZVUVIJVGH-UHFFFAOYSA-N trimethylxanthine Natural products CN1C(=O)N(C)C(=O)C2=C1N=CN2C RYYVLZVUVIJVGH-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
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A20/00—Water conservation; Efficient water supply; Efficient water use
- Y02A20/20—Controlling water pollution; Waste water treatment
- Y02A20/204—Keeping clear the surface of open water from oil spills
Landscapes
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of image processing, in particular to a method for detecting anti-wear hydraulic oil pollution particles. The method comprises the following steps: obtaining the gradient amplitude of a pixel point in each pixel block in the ferrographic image of the hydraulic oil and the contrast of each pixel block to obtain the texture similarity index of each pixel block and the pixel block adjacent to the pixel block, and further combining the pixel blocks to obtain each image block; obtaining each flow direction point pair in the window corresponding to each pixel point according to the gradient direction of the pixel point in the window corresponding to each pixel point; determining the selection necessity of each flow direction point pair according to the number of the flow direction point pairs in the window corresponding to each flow direction point pair and the edge extension line corresponding to each flow direction point pair, and further obtaining each target area; and obtaining an enhanced image according to the texture similarity indexes of each target area and the target areas adjacent to the target areas and the original gray level enhancement parameters of the target areas, and further judging whether the hydraulic oil contains pollution particles. The invention improves the detection precision of the hydraulic oil pollution particles.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a method for detecting anti-wear hydraulic oil pollution particles.
Background
In the mechanical movement of the hydraulic device, the pollution particles in the hydraulic oil often cause early wear of the joint surface of the friction pair of the hydraulic element, so that the moving part is damaged, or a valve port and a corrosion element are blocked, and the hydraulic device cannot normally work, therefore, the detection of the hydraulic oil is needed to judge whether the pollution particles exist in the hydraulic oil. A common image processing-based method is characterized in that various particles in an enhanced hydraulic oil ferrograph image are judged, and then whether pollution particles exist in hydraulic oil is determined, but when an acquired image is enhanced, magnetic particles and non-magnetic particles exist in the hydraulic oil pollution particles, and the non-magnetic particles are not affected by a magnetic field of an iron spectrometer and can be overlapped with the magnetic particles, so that the overall gray level enhancement effect of a region is poor, the gray level enhancement needs to be optimized, the existing multi-section linear graying has a poor overlapped edge enhancement effect close to the gray level, and further the detection accuracy of the hydraulic oil pollution particles is low.
Disclosure of Invention
In order to solve the problem of low detection precision when the existing method is used for detecting hydraulic oil pollution particles, the invention aims to provide an anti-wear hydraulic oil pollution particle detection method, which adopts the following technical scheme:
the invention provides a method for detecting anti-wear hydraulic oil pollution particles, which comprises the following steps:
acquiring an iron spectrum image of hydraulic oil to be detected;
dividing the ferrographic image to obtain pixel blocks, and obtaining texture similarity indexes of the pixel blocks and pixel blocks adjacent to the pixel blocks according to the gradient amplitude of the pixel points in the pixel blocks and the contrast corresponding to the pixel blocks; merging the pixel blocks based on the texture similarity indexes to obtain each image block; taking each pixel point in each image block as a center, and constructing a window corresponding to each pixel point; obtaining each flow direction point pair in the window corresponding to each pixel point according to the gradient direction of the pixel point in the window corresponding to each pixel point;
obtaining an edge extension line corresponding to each flow direction point pair based on the gradient direction of each pixel point in each flow direction point pair; determining the selection necessity of each flow direction point pair according to the number of the flow direction point pairs in the window corresponding to each flow direction point pair and the edge extension line corresponding to each flow direction point pair; obtaining each target area based on the selection necessity and the edge pixel points of each image block;
acquiring original gray level enhancement parameters of each target area, determining the target gray level enhancement parameters of each target area according to texture similarity indexes of each target area and the target areas adjacent to the target area and the original gray level enhancement parameters of each target area, and acquiring an enhanced image based on the target gray level enhancement parameters; and judging whether the hydraulic oil to be detected has pollution particles or not based on the enhanced image.
Preferably, the merging the pixel blocks based on the texture similarity index to obtain each image block includes:
repeatedly performing merging processing on each pixel block, wherein the merging processing comprises the following steps: judging whether the texture similarity index of each pixel block and each pixel block adjacent to the pixel block is larger than a threshold of the texture similarity index, if so, merging the two corresponding pixel blocks, judging whether the texture similarity index of the merged pixel block and each pixel block adjacent to the merged pixel block is larger than the threshold of the texture similarity index, if so, merging the corresponding pixel blocks, if not, reserving the original pixel block, and repeating the steps to mark each finally obtained merging area as an image block.
Preferably, the obtaining, according to the gradient direction of the pixel points in the window corresponding to each pixel point, each flow direction point pair in the window corresponding to each pixel point includes:
for the ith pixel point in any image block: and judging whether the gradient directions of two adjacent pixel points are opposite or not in a window corresponding to the ith pixel point, and if so, taking the corresponding two adjacent pixel points as a flow direction point pair.
Preferably, the determining the necessity of selecting each flow direction point pair according to the number of the flow direction point pairs in the window corresponding to the flow direction point pair and the edge extension line corresponding to the flow direction point pair includes:
for the jth flow direction point pair in any image block:
dividing an image block where a jth flow direction point pair is located into two sub-areas based on an edge extension line corresponding to the jth flow direction point pair, and marking texture similarity indexes of the two sub-areas divided by the jth flow direction point pair as first similarity indexes; counting the number of the flow direction point pairs in the window corresponding to the jth flow direction point pair; taking a natural constant as a base number, and taking the value of an exponential function with the first similarity index as an index as a first characteristic index; taking the reciprocal of the first characteristic index as the texture difference degree; and calculating the product of the number of the flow point pairs in the window corresponding to the jth flow point pair and the texture difference degree, and performing normalization processing to obtain a normalization result as the selection necessity of the jth flow point pair.
Preferably, the obtaining of the texture similarity index between each pixel block and the pixel block adjacent to the pixel block according to the gradient amplitude of the pixel point in each pixel block and the contrast corresponding to each pixel block includes:
calculating the texture similarity indexes of the qth pixel block and the adjacent pth pixel block by adopting the following formula:
Preferably, the obtaining of the edge extension line corresponding to each flow direction point pair based on the gradient direction of each pixel point in each flow direction point pair includes:
and respectively passing through the centers of the flow direction point pairs, making a straight line which is simultaneously perpendicular to the gradient direction of two pixel points in each flow direction point pair, and recording the straight line as the edge extension line corresponding to each flow direction point pair.
Preferably, the obtaining of each target region based on the selection necessity and the edge pixel point of each image block includes:
respectively judging whether the selection necessity of each flow direction point pair is greater than or equal to an necessity threshold, and if so, taking the center of the corresponding flow direction point pair as an edge pixel point of the target area; and obtaining the target area based on the edge pixel points of the target area.
Preferably, the determining the target gray scale enhancement parameter of each target region according to the texture similarity index of each target region and the target region adjacent to each target region, and the original gray scale enhancement parameter of each target region includes:
calculating the mean value of the texture similarity indexes of each target area and all target areas adjacent to the target area, recording the mean value as an average texture similarity index, carrying out positive correlation mapping on the average texture similarity index, and weighting the original gray level enhancement parameters of each target area based on the mapping result to obtain the target gray level enhancement parameters of each target area.
Preferably, the determining whether the hydraulic oil to be detected has the pollution particles based on the enhanced image includes:
inputting the enhanced image into a pre-trained neural network to obtain the corresponding category of each target area, wherein the category comprises pollution particles and non-pollution particles; and determining whether the pollution particles exist in the hydraulic oil to be detected or not based on the corresponding categories of the target areas.
The invention has at least the following beneficial effects:
1. the invention considers that magnetic particles and non-magnetic particles in hydraulic oil present different characteristics under the action of a magnetic field of an iron spectrometer, the non-magnetic particles are not influenced by the magnetic field of the iron spectrometer and can be overlapped with the magnetic particles, so that the overall gray scale enhancement effect of the region is poor, the existing multi-segment linear graying has poor enhancement effect on overlapped edges with similar gray values, the texture similarity indexes of each pixel block and pixel blocks adjacent to the pixel block are obtained according to the gradient amplitude and the corresponding contrast of the pixel blocks in an iron spectrum image of hydraulic oil to be detected, the larger the texture similarity indexes are, the higher the texture similarity degree of the two corresponding pixel blocks is, the more the two corresponding pixel blocks are subjected to merging processing, therefore, the invention obtains image blocks by merging the pixel blocks based on the texture similarity indexes, corrects the edges of the image blocks to obtain a target region, the obtaining result of the target region is more accurate, the detection precision of subsequent hydraulic oil pollution particles can be effectively ensured, the characteristics of the pixel blocks in each target region are similar, target gray scale enhancement parameters of different target regions are determined, the reinforced images are obtained, and the detection result that the detection of the multi-segment linear graying particle pollution of hydraulic oil pollution in the hydraulic oil to be detected can be directly classified according to the gray scale particle detection is better.
2. When the ferrographic image of the hydraulic oil to be detected is analyzed, the image is divided firstly, and then the division result is corrected, so that accurate identification of the particle area can be ensured on the basis of reducing the calculation amount of calculating pixel points one by one.
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 of a method for detecting anti-wear hydraulic oil pollution particles provided by the invention.
Detailed Description
In order to further illustrate the technical means and effects of the present invention adopted to achieve the predetermined objects, the following detailed description of a method for detecting anti-wear hydraulic oil contamination particles according to the present invention is provided with reference to the accompanying drawings and preferred 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.
The following specifically describes a specific scheme of the method for detecting the anti-wear hydraulic oil pollution particles provided by the invention with reference to the accompanying drawings.
The embodiment of a method for detecting anti-wear hydraulic oil pollution particles comprises the following steps:
the embodiment provides a method for detecting anti-wear hydraulic oil pollution particles, and as shown in fig. 1, the method for detecting anti-wear hydraulic oil pollution particles of the embodiment comprises the following steps:
s1, obtaining an iron spectrum image of hydraulic oil to be detected.
The specific scenario addressed by the present embodiment is as follows: collecting ferrographic images of hydraulic oil to be detected, analyzing the ferrographic images, detecting the ferrographic images of the hydraulic oil to be detected, wherein different particles may overlap with each other, so that the positions of the particles cannot be accurately divided by edge detection, dividing the collected ferrographic images to obtain a plurality of pixel blocks, calculating texture similarity indexes of two adjacent pixel blocks, judging whether adjacent image blocks need to be combined or not based on the texture similarity indexes, further obtaining a plurality of image blocks, correcting the edge of each image block to obtain an accurate target area, wherein the characteristics of the pixel points in the same target area are similar, enhancing different target areas in the images to different degrees based on the texture similarity indexes of each target area and the adjacent target area, and detecting wear-resistant hydraulic oil pollution particles based on the enhanced images.
Because the mechanical component is mostly a metal component, and the metal particles are mostly magnetic, the ferromagnetic wires can be arranged in the magnetic field of a ferrograph, but the nonmagnetic particles are, for example: aluminum, ceramic, external sand and stones and the like are not influenced by a magnetic field, are deposited in hydraulic oil and possibly overlap with magnetic particles in the hydraulic oil, so that the accuracy of particle division is influenced.
In the embodiment, an iron spectrum image of the antiwear hydraulic oil to be detected is collected through an iron spectrometer and a camera, then the collected image is subjected to graying, and the image subjected to graying is recorded as the iron spectrum image of the hydraulic oil to be detected and used for subsequent detection of the hydraulic oil pollution particles. Graying is prior art and will not be described in detail herein.
S2, dividing the ferrographic image to obtain pixel blocks, and obtaining texture similarity indexes of the pixel blocks and the pixel blocks adjacent to the pixel blocks according to the gradient amplitude of the pixel points in the pixel blocks and the contrast corresponding to the pixel blocks; merging the pixel blocks based on the texture similarity indexes to obtain each image block; taking each pixel point in each image block as a center, and constructing a window corresponding to each pixel point; and obtaining each flow direction point pair in the window corresponding to each pixel point according to the gradient direction of the pixel point in the window corresponding to each pixel point.
Because there may be the phenomenon of adhesion in some areas in the ferrographic image of the hydraulic oil that awaits measuring, if the gray scale based on the pixel in the ferrographic image of the hydraulic oil that awaits measuring directly carries out the pixel blocking, can make the adhesion edge distinguish unobviously, similar gray scale can not form great grey scale difference when multistage linear enhancement. In the embodiment, a ferrographic image of hydraulic oil to be detected is divided to obtain a plurality of pixel blocks, the gray scale relationship and the shape of different particles in the ferrographic image of hydraulic oil to be detected are judged, the pixel blocks are merged according to different attribution evaluation values, and the accurate edge after the pixel blocks are merged is judged, so that the gray scale enhancement in different degrees is performed on particle areas with different gray scale characteristics.
The generation principle of the non-magnetic particles is different from that of the magnetic particles, so that the textures of the region where the non-magnetic particles are located and the region where the magnetic particles are located are different, the magnetic particles are distributed near a ferromagnetic line under the action of a magnetic field of a ferrograph and are closer to the focusing position of a camera, clear and sharp internal textures are presented, the non-magnetic particles are deposited below, and the texture features of the magnetic particles are difficult to present in an image shot by the camera and are presented as fuzzy particle regions. Texture difference evaluation between blocks is judged through measurement of clear pixel textures contained in the pixel blocks, compared with the traditional method that division is directly carried out according to gray levels, calculation is carried out on the ferrographic image blocks, a large amount of redundancy existing in calculation of pixel points one by one is reduced, and meanwhile division accuracy of regions with anti-wear hydraulic oil pollution particles is guaranteed.
Specifically, in this embodiment, an iron spectrum image of hydraulic oil to be detected is first divided into pixel blocks with equal k × k areas, where the value of k is 16 in this embodiment, and in a specific application, an implementer may set the value of k according to the size of the iron spectrum image; on the basis of guaranteeing accurate identification of the particle region, the calculation amount of calculating pixel points one by one is reduced, the texture similarity degree between pixel blocks is evaluated, and the pixel blocks are conveniently and accurately divided.
After the image is partitioned, the more clear the texture inside each pixel block is, the higher the possibility that the pixel block is a magnetic particle area is, the larger the difference of the texture among the pixel blocks is, and the more dissimilar the texture is; if two adjacent pixel blocks are more similar, it indicates that the corresponding two pixel blocks are more suitable for performing the merging process, so the following embodiment analyzes the similarity of each two adjacent pixel blocks to determine whether to perform the merging process. Firstly, acquiring the gradient amplitude and the gradient direction of each pixel point in an iron spectrum image of hydraulic oil to be detected by using a Sobel operator; considering that the characteristics between the pixel points are reflected only by the gradient amplitude difference value, the gray difference between the whole current pixel block and other pixel blocks cannot be judged, the contrast of the pixel blocks needs to be synthesized for judgment, the gray difference between different pixel blocks is reflected by using the contrast, and the whole characteristics of the pixel blocks are reflected, so that the contrast of each pixel block in the ferrographic image of the hydraulic oil to be detected is calculated, and the gradient amplitude, the gradient direction and the contrast acquisition method are the prior art and are not described in detail herein; calculating texture similarity indexes of every two adjacent pixel blocks based on the gradient amplitude of pixel points in every two adjacent pixel blocks and the contrast of every pixel block; for the qth pixel block and the pth pixel block adjacent to the qth pixel block, the calculation formula of the texture similarity index of the two pixel blocks is as follows:
And traversing each pixel block in the ferrographic image of the hydraulic oil to be detected to obtain the texture similarity index of each pixel block and the pixel block adjacent to the pixel block in the ferrographic image of the hydraulic oil to be detected.
The larger the texture similarity index is, the more similar the textures of the two corresponding pixel blocks are, and since the pixel blocks are divided based on a fixed size, a region of the same category may be divided into a plurality of pixel blocks, and therefore, it is necessary to determine whether or not the two adjacent pixel blocks need to be merged based on the texture similarity index. Specifically, a texture similarity index threshold value is set
In this embodiment, the
In a specific application, the implementer can set the setting according to specific situations. Judging the texture similarity index of the qth pixel block and the adjacent pth pixel block
Whether or not greater than
If the difference is greater than the threshold value, combining the qth pixel block and the pth pixel block adjacent to the qth pixel block to obtain a combined pixel block, and similarly, calculating texture similarity indexes of the combined pixel block and the pixel block adjacent to the combined pixel block by adopting the method, and based on the texture similarity indexes and the pixel blocks adjacent to the combined pixel block
Whether the combination is continued or not is judged until the texture similarity indexes of the combined pixel block and all the adjacent pixel blocks are less than or equal to
(ii) a If the number of pixels is less than or equal to the q-th pixel block, the q-th pixel block and the adjacent q-th pixel block are explainedThe texture similarity of the p pixel blocks is low, and the p pixel blocks are not combined. And processing other pixel blocks in the ferrographic image of the hydraulic oil to be detected by adopting the method, traversing from the origin of the image coordinate system along the sequence of the first row and the second row of the image, marking each pixel block finally combined in the ferrographic image of the hydraulic oil to be detected as an image block, namely dividing the ferrographic image of the hydraulic oil to be detected into a plurality of image blocks.
Most of hydraulic oil mechanical components are steel and iron parts with magnetism, harmful particles are mainly abrasion particles generated by metal fatigue of machine parts and a small amount of nonmagnetic externally-invaded sand particles, and hydraulic oil generates oxidation pollution due to use, so that the oil pollution with uneven color is generated except the particles. Therefore, different degrees of enhancement are required for different image blocks.
In this embodiment, pixel blocks are synthesized according to the texture similarity index of adjacent pixel blocks to obtain a plurality of image blocks. Considering that the edge of an image block obtained by synthesizing a pixel block is formed by a broken line, and the edge of the same image block may contain pixel points on particles of other types, the edge of the image block needs to be corrected to obtain an accurate division result, so that the result of particle detection after image enhancement is more accurate.
Because there is grey level difference at the granule edge, consequently can detect the granule edge, but consider that there is the difference in the material of granule and lead to that the pixel gradient direction is comparatively disorderly in the granule place region, can not directly obtain the edge through current edge detection, need combine the gradient direction of pixel to carry out the pertinence to the marginal pixel and select. The multi-section gray scale linear enhancement endows different gray scale levels with different gray scale linear enhancement coefficients, facilitates the stretching and compression of different gray scale levels, improves the image quality, highlights the image details and facilitates the judgment of the specific types and the generation reasons of particles in the ferrographic image.
The gradient direction of each pixel point in the ferrographic image of the hydraulic oil to be detected is already obtained, and then the gradient direction of the pixel points is analyzed, so that the edge line of the divided image block is corrected.
For the ith pixel point in any image block: taking the ith pixel point as a window center, constructing a window with a preset size as a window corresponding to the ith pixel point; in this embodiment, the preset size is 5 × 5, and in specific applications, an implementer may set the size according to specific situations; judging whether gradient directions of two adjacent pixel points are opposite or not in a window corresponding to the ith pixel point, wherein an included angle formed by the gradient directions of the two adjacent pixel points is required to be 180 degrees, if so, indicating that a gray edge exists between the two corresponding pixel points, taking the two corresponding adjacent pixel points as a flow direction point pair, and counting the number of the flow direction point pairs in the window corresponding to the ith pixel point. By adopting the method, the flow direction point pairs and the number of the flow direction point pairs in the window corresponding to each pixel point in each image block can be obtained.
S3, obtaining an edge extension line corresponding to each flow direction point pair based on the gradient direction of each pixel point in each flow direction point pair; determining the selection necessity of each flow direction point pair according to the number of the flow direction point pairs in the window corresponding to each flow direction point pair and the edge extension line corresponding to each flow direction point pair; and obtaining each target area based on the selection necessity and the edge pixel points of each image block.
Because multiple pairs of flow direction points may exist in a window corresponding to the same pixel point, which causes an uncertain edge direction, it is necessary to construct an area edge by combining the similar indexes of the internal and external textures of the divided edge with the flow direction points, and to eliminate the influence of the false edge inside the area.
For any image block:
for any flow direction point pair in the image block, because the included angle formed by the gradient directions of two pixel points in the flow direction point pair is 180 degrees, a straight line which is simultaneously perpendicular to the gradient directions of the two pixel points in the flow direction point pair is made through the center of the flow direction point pair, the straight line is marked as an edge extension line corresponding to the flow direction point pair, the image block is divided into two areas by the edge extension line corresponding to the flow direction point pair, the two areas are respectively marked as two sub-areas, the more dissimilar the textures of the two sub-areas are, and the more accurate the edge division is illustrated. After the flow direction point pair is divided, the obtained textures of the two sub-regions are more dissimilar, which indicates that the difference between the two regions is increased after the current flow direction point pair is divided, and the probability that the edge divided by the current flow direction point pair is the real particle edge is higher. The texture similarity index of the sub-region obtained after division is used for performing similarity evaluation on the surrounding region of the flow direction point pair, namely, part of characteristics are evaluated, and the noise margin inside some particles is more, so that the judgment result only through texture similarity is influenced. Specifically, in this embodiment, the necessity of selecting each flow direction point pair is determined according to the number of the flow direction point pairs in the window corresponding to each flow direction point pair and the texture similarity index of the two sub-regions divided by the edge extension line corresponding to each flow direction point pair; for the jth flow direction point pair in the image block, marking texture similarity indexes of two sub-areas divided by the jth flow direction point pair as first similarity indexes, counting the number of the flow direction point pairs in a window corresponding to the jth flow direction point pair, taking a natural constant as a base number, taking a value of an exponential function with the first similarity index as an index as a first characteristic index, taking the reciprocal of the first characteristic index as the texture difference degree, calculating the product of the number of the flow direction point pairs in the window corresponding to the jth flow direction point pair and the texture difference degree, carrying out normalization processing, and taking a normalization result as the selection necessity of the jth flow direction point pair; the specific calculation formula of the necessity of selecting the jth flow direction point pair in the window corresponding to the ith pixel point in the image block is as follows:
wherein ,
the necessity of selecting the jth flow direction point pair in the window corresponding to the ith pixel point in the image block,
the number of pairs of flow-direction points in the window corresponding to the ith pixel point in the image block,
the texture similarity index of the two sub-areas divided from the jth flow direction point pair in the window corresponding to the ith pixel point in the image block,
in order to be a function of the normalization,
is a natural constant.
By adopting the method, the selection necessity of each flow direction point pair in each image block can be obtained, and the necessity threshold value is set
In this embodiment
The value of (A) is 0.9, which can be set by the practitioner according to the particular situation in the particular application; respectively judging whether the selection necessity of each flow direction point pair is more than or equal to
If the current point pair is not less than the preset value, the corresponding current point pair is suitable for obtaining the edge pixel point of the target area, and the center of the corresponding current point pair is used as the edge pixel point of the target area; if the current point pair is smaller than the target area, the corresponding current point pair is not suitable for obtaining the edge pixel point of the target area. By adopting the method, the edge pixel points of all target areas in the ferrographic image of the hydraulic oil to be detected are obtained, the ferrographic image of the hydraulic oil to be detected is detected through the existing connected domain detection algorithm based on the edge pixel points of all target areas in the ferrographic image of the hydraulic oil to be detected, the target areas in the ferrographic image of the hydraulic oil to be detected are obtained, and the ferrographic image of the hydraulic oil to be detected is divided into a plurality of target areas.
S4, acquiring original gray level enhancement parameters of each target area, determining the target gray level enhancement parameters of each target area according to texture similarity indexes of each target area and the target areas adjacent to the target area and the original gray level enhancement parameters of each target area, and acquiring an enhanced image based on the target gray level enhancement parameters; and judging whether the hydraulic oil to be detected has pollution particles or not based on the enhanced image.
And each target area in the ferrographic image of the hydraulic oil to be detected has an adjacent target area, and the gray scale enhancement linear parameter is calculated through the texture similarity index between the two adjacent target areas, so that the ferrographic image of the hydraulic oil to be detected is enhanced. The method for calculating the texture similarity index between two adjacent target regions is the same as the method for calculating the texture similarity index of two adjacent pixel blocks in step S2, and is not described in detail here. The smaller the texture similarity index between two adjacent target areas is, the larger the gray level difference between the two corresponding target areas is, and the smaller the gray level enhancement degree required by the corresponding target area is.
For the a-th target area:
acquiring original gray level enhancement parameters of the target area, calculating the mean value of the texture similar indexes of the target area and all target areas adjacent to the target area according to the texture similar indexes of the target area and each target area adjacent to the target area, recording the mean value as an average texture similar index, performing positive correlation mapping on the average texture similar index, weighting the original gray level enhancement parameters of the target area based on the mapping result to obtain target gray level enhancement parameters of the target area, and enhancing the target area based on the target gray level enhancement parameters of the target area; the method for acquiring the original gray level enhancement parameters is the prior art, and redundant description is omitted here; the specific expression of the target gray level enhancement parameter of the target area is as follows:
, wherein ,
is the target gray scale enhancement parameter for the target region,
for the original grayscale enhancement parameter of the target region,
the average value of the texture similarity indexes of the target area and all the target areas adjacent to the target area is the average value of the texture similarity indexes, namely the average texture similarity index.
By adopting the method, the target gray level enhancement parameter of each target area in the ferrographic image of the hydraulic oil to be detected can be obtained, and the corresponding target area is enhanced based on the target gray level enhancement parameter of each target area, namely, the enhanced image is obtained for the subsequent detection of the anti-friction hydraulic oil pollution particles. Compared with the traditional method of carrying out multi-section gray scale linear enhancement on an image according to a fixed gray scale range, the embodiment divides the target region through the texture similarity index of the target region, calculates the target gray scale enhancement parameter of each target region according to the gray scale difference between the target regions, reduces the problem that calculation amount is large when pixel points are calculated one by one, and simultaneously carries out enhancement processing on the whole of each target region, thereby avoiding that the change of different gray scales of different particles in a ferrograph image in the fixed gray scale range can not reach better characteristic of the ROI region to be more remarkable, and leading different particles in the image to be capable of being subjected to targeted enhancement processing.
In the embodiment, a neural network is used for detecting an enhanced image and judging whether pollution particles exist in hydraulic oil to be detected, the neural network in the embodiment adopts an Encoder-Decoder structure, the enhanced image is input into the neural network trained in advance, and the category corresponding to each target area in the enhanced image is output, wherein the category comprises the pollution particles and non-pollution particles; the training process of the neural network is a known technology, and redundant description is not repeated here; by adopting the method provided by the embodiment, whether the pollution particles exist in the hydraulic oil to be detected is accurately detected, and in specific application, an implementer can evaluate the pollution of the hydraulic oil to be detected according to the output result of the neural network.
In the embodiment, the magnetic particles and the non-magnetic particles in the hydraulic oil are different in characteristics under the action of a magnetic field of an iron spectrometer, the non-magnetic particles are not affected by the magnetic field of the iron spectrometer and can be overlapped with the magnetic particles, so that the overall gray scale enhancement effect of the region is poor, the enhancement effect of the existing multi-segment linear graying on overlapped edges with similar gray values is poor, according to the gradient amplitude and the corresponding contrast of pixel points in pixel blocks in an iron spectrum image of the hydraulic oil to be detected, texture similarity indexes of the pixel blocks and pixel blocks adjacent to the pixel blocks are obtained, the larger the texture similarity index is, the higher the texture similarity degree of the corresponding two pixel blocks is, the more the two pixel blocks are subjected to merging processing, therefore, the embodiment performs merging processing on the pixel blocks based on the texture similarity indexes to obtain image blocks, corrects the edges of the image blocks to obtain target regions, the characteristics of the pixel points in each target region are similar, determines target gray scale enhancement parameters of different target regions, obtains an enhanced image, further judges whether the hydraulic oil to be detected, and can effectively detect the pollution particles of the hydraulic oil to the multi-segment linear graying images, and can overcome the defect that the pollution of the iron spectrum image to be detected. When the ferrographic image of the hydraulic oil to be detected is analyzed, the image is divided firstly, then the division result is corrected, and accurate identification of the particle area can be ensured on the basis of reducing the calculation amount of calculating pixel points one by one.
Claims (9)
1. The method for detecting the anti-wear hydraulic oil pollution particles is characterized by comprising the following steps of:
acquiring an iron spectrum image of hydraulic oil to be detected;
dividing the ferrographic image to obtain pixel blocks, and obtaining texture similarity indexes of the pixel blocks and pixel blocks adjacent to the pixel blocks according to the gradient amplitude of the pixel points in the pixel blocks and the contrast corresponding to the pixel blocks; merging the pixel blocks based on the texture similarity indexes to obtain each image block; taking each pixel point in each image block as a center, and constructing a window corresponding to each pixel point; obtaining each flow direction point pair in the window corresponding to each pixel point according to the gradient direction of the pixel point in the window corresponding to each pixel point;
obtaining an edge extension line corresponding to each flow direction point pair based on the gradient direction of each pixel point in each flow direction point pair; determining the selection necessity of each flow direction point pair according to the number of the flow direction point pairs in the window corresponding to each flow direction point pair and the edge extension line corresponding to each flow direction point pair; obtaining each target area based on the selection necessity and the edge pixel points of each image block;
acquiring original gray level enhancement parameters of each target area, determining target gray level enhancement parameters of each target area according to texture similarity indexes of each target area and the target areas adjacent to the target area and the original gray level enhancement parameters of each target area, and acquiring an enhanced image based on the target gray level enhancement parameters; and judging whether the hydraulic oil to be detected has pollution particles or not based on the enhanced image.
2. The method for detecting anti-wear hydraulic oil pollution particles according to claim 1, wherein the merging of pixel blocks based on the texture similarity index to obtain each image block comprises:
repeatedly performing merging processing on each pixel block, wherein the merging processing comprises the following steps: judging whether the texture similarity index of each pixel block and each pixel block adjacent to the pixel block is larger than a threshold value of the texture similarity index, if so, merging the two corresponding pixel blocks, judging whether the texture similarity index of the merged pixel block and each pixel block adjacent to the merged pixel block is larger than the threshold value of the texture similarity index, if so, merging the corresponding pixel blocks, if not, reserving the original pixel block, and so on, and finally marking each obtained merging area as an image block.
3. The method for detecting the anti-wear hydraulic oil pollution particles according to claim 1, wherein the step of obtaining the flow direction point pairs in the window corresponding to each pixel point according to the gradient direction of the pixel points in the window corresponding to each pixel point comprises the following steps:
for the ith pixel point in any image block: and judging whether the gradient directions of two adjacent pixel points are opposite or not in a window corresponding to the ith pixel point, and if so, taking the corresponding two adjacent pixel points as a flow direction point pair.
4. The method for detecting anti-wear hydraulic oil contamination particles according to claim 1, wherein the determining the necessity of selecting each flow direction point pair according to the number of the flow direction point pairs in the window corresponding to each flow direction point pair and the edge extension line corresponding to each flow direction point pair comprises:
for the jth flow direction point pair in any image block:
dividing an image block where a jth flow direction point pair is located into two sub-areas based on an edge extension line corresponding to the jth flow direction point pair, and marking texture similarity indexes of the two sub-areas divided by the jth flow direction point pair as first similarity indexes; counting the number of the flow direction point pairs in the window corresponding to the jth flow direction point pair; taking a natural constant as a base number, and taking the value of an exponential function with the first similarity index as an index as a first characteristic index; taking the reciprocal of the first characteristic index as the texture difference degree; and calculating the product of the number of the flow point pairs in the window corresponding to the jth flow point pair and the texture difference degree, and performing normalization processing to obtain a normalization result as the selection necessity of the jth flow point pair.
5. The method for detecting anti-wear hydraulic oil pollution particles according to claim 1, wherein the step of obtaining texture similarity indexes of each pixel block and pixel blocks adjacent to the pixel block according to the gradient amplitude of the pixel points in each pixel block and the contrast corresponding to each pixel block comprises the steps of:
calculating the texture similarity indexes of the qth pixel block and the pth pixel block adjacent to the qth pixel block by adopting the following formula:
6. The method for detecting anti-wear hydraulic oil pollution particles according to claim 1, wherein obtaining the edge extension line corresponding to each flow direction point pair based on the gradient direction of each pixel point in each flow direction point pair comprises:
and respectively passing through the center of each flow direction point pair, making a straight line which is simultaneously perpendicular to the gradient direction of two pixel points in each flow direction point pair, and recording the straight line as the edge extension line corresponding to each flow direction point pair.
7. The method for detecting anti-wear hydraulic oil pollution particles according to claim 1, wherein the obtaining of each target area based on the selection necessity and the edge pixel points of each image block comprises:
respectively judging whether the selection necessity of each flow direction point pair is greater than or equal to a necessity threshold, and if so, taking the center of the corresponding flow direction point pair as an edge pixel point of the target area; and obtaining the target area based on the edge pixel points of the target area.
8. The method for detecting anti-wear hydraulic oil pollution particles according to claim 1, wherein the determining the target gray scale enhancement parameter of each target area according to the texture similarity index of each target area and the target area adjacent to the target area and the original gray scale enhancement parameter of each target area comprises:
calculating the mean value of the texture similarity indexes of each target area and all target areas adjacent to the target area, recording the mean value as an average texture similarity index, carrying out positive correlation mapping on the average texture similarity index, and weighting the original gray level enhancement parameters of each target area based on the mapping result to obtain the target gray level enhancement parameters of each target area.
9. The method for detecting anti-wear hydraulic oil pollution particles according to claim 1, wherein the step of judging whether the pollution particles exist in the hydraulic oil to be detected based on the enhanced image comprises the steps of:
inputting the enhanced image into a pre-trained neural network to obtain the corresponding category of each target area, wherein the category comprises pollution particles and non-pollution particles; and determining whether the pollution particles exist in the hydraulic oil to be detected or not based on the corresponding categories of the target areas.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310252093.6A CN115965624B (en) | 2023-03-16 | 2023-03-16 | Method for detecting wear-resistant hydraulic oil pollution particles |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310252093.6A CN115965624B (en) | 2023-03-16 | 2023-03-16 | Method for detecting wear-resistant hydraulic oil pollution particles |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115965624A true CN115965624A (en) | 2023-04-14 |
| CN115965624B CN115965624B (en) | 2023-05-30 |
Family
ID=85905178
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310252093.6A Active CN115965624B (en) | 2023-03-16 | 2023-03-16 | Method for detecting wear-resistant hydraulic oil pollution particles |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115965624B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116542883A (en) * | 2023-07-07 | 2023-08-04 | 四川大学华西医院 | Magnetic control capsule gastroscope image focus mucosa enhancement system |
| CN116645368A (en) * | 2023-07-27 | 2023-08-25 | 青岛伟东包装有限公司 | Online visual detection method for edge curl of casting film |
| CN116823811A (en) * | 2023-08-25 | 2023-09-29 | 汶上县誉诚制衣有限公司 | Functional jacket surface quality detection method |
| CN117455913A (en) * | 2023-12-25 | 2024-01-26 | 卡松科技股份有限公司 | Intelligent detection method for pollution of hydraulic oil based on image features |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110223299A (en) * | 2019-06-12 | 2019-09-10 | 重庆邮电大学 | A kind of particle partition method based on deposition process |
| CN112667717A (en) * | 2020-12-23 | 2021-04-16 | 贵州电网有限责任公司电力科学研究院 | Transformer substation inspection information processing method and device, computer equipment and storage medium |
| CN114943739A (en) * | 2022-07-26 | 2022-08-26 | 山东三微新材料有限公司 | Aluminum pipe quality detection method |
| CN115063409A (en) * | 2022-07-28 | 2022-09-16 | 南通恒强轧辊有限公司 | Method and system for detecting surface material of mechanical cutter |
| CN115115638A (en) * | 2022-08-30 | 2022-09-27 | 南通翡利达液压科技有限公司 | Oil leakage detection and judgment method for hydraulic system |
| WO2022237811A1 (en) * | 2021-05-11 | 2022-11-17 | 北京字跳网络技术有限公司 | Image processing method and apparatus, and device |
| CN115358947A (en) * | 2022-08-22 | 2022-11-18 | 西安交通大学 | Ferrographic abrasive particle image denoising method, device and equipment and readable storage medium |
| CN115546127A (en) * | 2022-09-21 | 2022-12-30 | 厦门大学 | Engine lubricating oil abrasive particle image analysis method and device, medium and equipment |
-
2023
- 2023-03-16 CN CN202310252093.6A patent/CN115965624B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110223299A (en) * | 2019-06-12 | 2019-09-10 | 重庆邮电大学 | A kind of particle partition method based on deposition process |
| CN112667717A (en) * | 2020-12-23 | 2021-04-16 | 贵州电网有限责任公司电力科学研究院 | Transformer substation inspection information processing method and device, computer equipment and storage medium |
| WO2022237811A1 (en) * | 2021-05-11 | 2022-11-17 | 北京字跳网络技术有限公司 | Image processing method and apparatus, and device |
| CN114943739A (en) * | 2022-07-26 | 2022-08-26 | 山东三微新材料有限公司 | Aluminum pipe quality detection method |
| CN115063409A (en) * | 2022-07-28 | 2022-09-16 | 南通恒强轧辊有限公司 | Method and system for detecting surface material of mechanical cutter |
| CN115358947A (en) * | 2022-08-22 | 2022-11-18 | 西安交通大学 | Ferrographic abrasive particle image denoising method, device and equipment and readable storage medium |
| CN115115638A (en) * | 2022-08-30 | 2022-09-27 | 南通翡利达液压科技有限公司 | Oil leakage detection and judgment method for hydraulic system |
| CN115546127A (en) * | 2022-09-21 | 2022-12-30 | 厦门大学 | Engine lubricating oil abrasive particle image analysis method and device, medium and equipment |
Non-Patent Citations (4)
| Title |
|---|
| SONG FENG ET AL.: "A Wear Debris Segmentation Method for Direct Reflection Online Visual Ferrography", 《SENSORS》, pages 1 - 12 * |
| T. B. KIRK ET AL.: "F’ractal parameters and computer image analysis applied to wear particles isolated by ferrography", 《ELSEVIER SEQUOIA/PRINTED IN THE NETHERLAND》, pages 347 - 365 * |
| 杨进: "液压油固体颗粒污染检测技术与方法", 《淮海工学院学报》, pages 14 - 16 * |
| 詹松 等: "磨损磨粒铁谱分析中图像预处理技术研究", 《合肥工业大学学报（自然科学版）》, vol. 27, no. 1, pages 44 - 47 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116542883A (en) * | 2023-07-07 | 2023-08-04 | 四川大学华西医院 | Magnetic control capsule gastroscope image focus mucosa enhancement system |
| CN116542883B (en) * | 2023-07-07 | 2023-09-05 | 四川大学华西医院 | Magnetic control capsule gastroscope image focus mucosa enhancement system |
| CN116645368A (en) * | 2023-07-27 | 2023-08-25 | 青岛伟东包装有限公司 | Online visual detection method for edge curl of casting film |
| CN116645368B (en) * | 2023-07-27 | 2023-10-03 | 青岛伟东包装有限公司 | Online visual detection method for edge curl of casting film |
| CN116823811A (en) * | 2023-08-25 | 2023-09-29 | 汶上县誉诚制衣有限公司 | Functional jacket surface quality detection method |
| CN116823811B (en) * | 2023-08-25 | 2023-12-01 | 汶上县誉诚制衣有限公司 | Functional jacket surface quality detection method |
| CN117455913A (en) * | 2023-12-25 | 2024-01-26 | 卡松科技股份有限公司 | Intelligent detection method for pollution of hydraulic oil based on image features |
| CN117455913B (en) * | 2023-12-25 | 2024-03-08 | 卡松科技股份有限公司 | Intelligent detection method for pollution of hydraulic oil based on image features |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115965624B (en) | 2023-05-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115965624A (en) | Detection method for anti-wear hydraulic oil pollution particles | |
| CN109060836B (en) | Machine vision-based high-pressure oil pipe joint external thread detection method | |
| CN115330783A (en) | Steel wire rope defect detection method | |
| CN115082482A (en) | Metal surface defect detection method | |
| CN109434251B (en) | Welding seam image tracking method based on particle filtering | |
| CN115661147B (en) | Metering detection data identification method based on machine vision | |
| CN104867130B (en) | A kind of self-adapting division method based on crack image sub-district gray average | |
| CN116385450B (en) | PS sheet wear resistance detection method based on image processing | |
| CN116137036B (en) | Gene detection data intelligent processing system based on machine learning | |
| CN115311629B (en) | Abnormal bending precision monitoring system of bending machine | |
| CN116664569B (en) | Weld flash defect detection method | |
| CN115294159B (en) | Method for dividing corroded area of metal fastener | |
| CN112580476B (en) | Sperm identification and multi-target track tracking method | |
| CN111652790A (en) | Sub-pixel image registration method | |
| CN107516315B (en) | Tunneling machine slag tapping monitoring method based on machine vision | |
| CN110458785B (en) | Magnetic levitation ball levitation gap detection method based on image sensing | |
| CN115908362A (en) | Method for detecting wear resistance of skateboard wheel | |
| CN110334727B (en) | Intelligent matching detection method for tunnel cracks | |
| CN103968772A (en) | Piston ring detection method based on monocular vision detection | |
| CN115239661A (en) | Mechanical part burr detection method and system based on image processing | |
| CN112408153B (en) | Method for monitoring movement amount of steel wire rope in elevator braking process | |
| CN116110006B (en) | Scenic spot tourist abnormal behavior identification method for intelligent tourism system | |
| CN107369163B (en) | Rapid SAR image target detection method based on optimal entropy dual-threshold segmentation | |
| CN116977230A (en) | Scanning electron microscope image optimization enhancement method | |
| Li et al. | Vehicle seat detection based on improved RANSAC-SURF algorithm |
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 | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: A method for detecting particles of anti-wear hydraulic oil contamination Effective date of registration: 20231215 Granted publication date: 20230530 Pledgee: China Postal Savings Bank Co.,Ltd. Jining Rencheng District sub branch Pledgor: Shandong Yuchi New Material Technology Co.,Ltd. Registration number: Y2023980071880 |
|
| PE01 | Entry into force of the registration of the contract for pledge of patent right |