CN115311276A - Intelligent segmentation method for ferrographic image based on machine vision - Google Patents

Intelligent segmentation method for ferrographic image based on machine vision Download PDF

Info

Publication number
CN115311276A
CN115311276A CN202211241929.4A CN202211241929A CN115311276A CN 115311276 A CN115311276 A CN 115311276A CN 202211241929 A CN202211241929 A CN 202211241929A CN 115311276 A CN115311276 A CN 115311276A
Authority
CN
China
Prior art keywords
color
points
initial
category
obtaining
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202211241929.4A
Other languages
Chinese (zh)
Other versions
CN115311276B (en
Inventor
汤维银
孙亚林
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jiangsu Huawei Photoelectric Technology Co ltd
Original Assignee
Jiangsu Huawei Photoelectric Technology Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Jiangsu Huawei Photoelectric Technology Co ltd filed Critical Jiangsu Huawei Photoelectric Technology Co ltd
Priority to CN202211241929.4A priority Critical patent/CN115311276B/en
Publication of CN115311276A publication Critical patent/CN115311276A/en
Application granted granted Critical
Publication of CN115311276B publication Critical patent/CN115311276B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0004—Industrial image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/90—Determination of colour characteristics
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10056—Microscopic image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30108—Industrial image inspection
    • G06T2207/30136—Metal

Landscapes

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

Abstract

The invention relates to the field of machine vision, in particular to a ferrographic image intelligent segmentation method based on machine vision, which comprises the following steps: converting all pixel points in the ferrographic image into color points in a Lab color space; obtaining all initial categories of all color points according to the initial clustering center and the first color difference value; obtaining the appropriateness of each target color point in each initial category, obtaining the appropriateness of all color points in each initial category according to an appropriateness prediction formula, and further obtaining a new clustering center; obtaining the final categories of all color points through multiple iterations; and obtaining a segmentation result of the ferrographic image according to the final classification of all the color points. The method of the invention enables the segmentation result of the ferrographic image to better accord with the expected segmentation effect of human vision, and provides better basis for the wear condition monitoring and fault diagnosis of equipment.

Description

Intelligent segmentation method for ferrographic image based on machine vision
Technical Field
The invention relates to the field of machine vision, in particular to a ferrographic image intelligent segmentation method based on machine vision.
Background
The ferrography technology is a technology for analyzing the characteristics of the abrasive dust such as granularity, shape, color, texture and the like, and can be used for wear condition monitoring and fault diagnosis. If the equipment is worn, the main wear state of the current equipment can be analyzed and obtained through the common characteristic that the oil liquid of the equipment contains the abrasive dust.
When the characteristics of the abrasive dust are extracted, the abrasive dust needs to be segmented, in the existing abrasive dust segmentation method, iron spectrum images of the abrasive dust are clustered based on color difference, certain areas in the iron spectrum images of the abrasive dust are segmented into a connected domain under the vision of human eyes, but under the existing abrasive dust segmentation method, the iron spectrum images of the abrasive dust can be segmented into different connected domains, so that the segmentation effect of the iron spectrum images of the abrasive dust is poor, and further the monitoring of abrasion working conditions and fault diagnosis are influenced.
Therefore, the scheme provides the intelligent iron spectrum image segmentation method based on machine vision, improves the segmentation effect of the iron spectrum image of the abrasive dust based on color segmentation, and provides a better basis for wear condition monitoring and fault diagnosis of equipment.
Disclosure of Invention
In order to solve the above problems, the present invention provides a ferrographic image intelligent segmentation method based on machine vision, the method comprising:
obtaining an iron spectrum image of abrasive dust in the oil sample through a microscopic module; converting all pixel points in the ferrographic image into color points in a Lab color space;
randomly selecting a first number of color points from all color points in the Lab color space, and recording the color points as initial clustering centers;
s1: obtaining all initial classes of all color points according to all initial cluster centers, including:
calculating all first color difference values of all color points in all initial clustering centers and Lab color spaces; clustering all the color points by using a K-means clustering algorithm according to all the first color difference values and the initial clustering centers to obtain all initial categories of all the color points;
s2, obtaining all new clustering centers according to all initial categories, including:
for any one initial category, randomly sampling a plurality of color points from all the color points of the initial category, and marking as target color points; calculating all second color difference values of all target color points and all other color points in the initial category, and obtaining the appropriate degree of the target color points according to all the second color difference values; calculating the appropriateness of each color point of the initial class according to the appropriateness of all the target color points; recording the color point with the maximum suitable degree in all the color points of the initial category as a new clustering center; for all initial categories, obtaining all new clustering centers;
repeatedly executing S1 and S2 by taking the new clustering center as an initial clustering center until the initial clustering center is not changed any more, and taking the obtained initial category as a final category;
and clustering all pixel points in the ferrograph image according to all color points corresponding to the final category to obtain the segmentation result of all pixel points in the ferrograph image.
Further, the step of the first color difference value and the second color difference value comprises:
obtaining the tolerance of the initial clustering center, judging whether the color point in the Lab color space is within the tolerance of the initial clustering center or not, and if the color point in the Lab color space is within the tolerance of the initial clustering center, determining that the first hue difference value is
Figure DEST_PATH_IMAGE001
(ii) a Otherwise, the first color difference value is:
Figure DEST_PATH_IMAGE003
wherein the content of the first and second substances,
Figure 479032DEST_PATH_IMAGE004
is the luminance value of the initial cluster center,
Figure DEST_PATH_IMAGE005
is the brightness value of a color point in the Lab color space,
Figure 176730DEST_PATH_IMAGE006
is the color value of the initial cluster center,
Figure DEST_PATH_IMAGE007
is the color value of a color point in the Lab color space,
Figure 129642DEST_PATH_IMAGE008
is a first hue difference value;
obtaining the tolerance of the target color point, judging whether other color points in the initial category are within the tolerance of the target color point, and if the other color points in the initial category are within the tolerance of the target color point, determining that the second color difference value is
Figure DEST_PATH_IMAGE009
(ii) a Otherwise the second hue value is:
Figure DEST_PATH_IMAGE011
wherein the content of the first and second substances,
Figure 284287DEST_PATH_IMAGE012
for the luminance value of the point of the target color,
Figure DEST_PATH_IMAGE013
the luminance values of the other color points in the initial class,
Figure 955440DEST_PATH_IMAGE014
is the color value of the target color point,
Figure 660090DEST_PATH_IMAGE014
the color values of the other color points in the initial category,
Figure DEST_PATH_IMAGE015
is the second hue value.
Further, the step of clustering all the color points by using a K-means clustering algorithm according to all the first color difference values and the initial clustering centers to obtain all the initial categories of all the color points includes:
according to the first color difference value of the color point and each initial clustering center, based on the minimum color difference value principle of the K-means clustering algorithm, the color point is divided into the initial clustering centers corresponding to the minimum first color difference value, a set formed by all the color points corresponding to each initial clustering center is marked as an initial category, and all the initial categories of all the color points are obtained.
Further, the step of obtaining the suitability degree of the target color point according to all the second color difference values comprises:
recording a sequence formed by all second color difference values of the target color point and all other color points in the initial category as a color difference value sequence of the target color point, and obtaining the appropriateness degree of the target color point according to the color difference value sequence of the target color point, wherein the calculation formula of the appropriateness degree of the target color point is as follows:
Figure DEST_PATH_IMAGE017
wherein the content of the first and second substances,
Figure 529826DEST_PATH_IMAGE018
an exponential function with a natural constant as the base is represented,
Figure DEST_PATH_IMAGE019
is as follows
Figure DEST_PATH_IMAGE021
In a category of
Figure DEST_PATH_IMAGE023
The variance of all color difference values in the sequence of color difference values for each target color point,
Figure 389460DEST_PATH_IMAGE024
is as follows
Figure 16751DEST_PATH_IMAGE021
In a category of
Figure 259513DEST_PATH_IMAGE023
All colors in the sequence of color difference values of the individual target color pointsThe difference between the maximum value and the minimum value of the difference,
Figure DEST_PATH_IMAGE025
is as follows
Figure 187018DEST_PATH_IMAGE021
In a category of
Figure 355569DEST_PATH_IMAGE023
The mean of all color difference values in the sequence of color difference values for the individual target color points,
Figure 470155DEST_PATH_IMAGE026
is as follows
Figure 516609DEST_PATH_IMAGE021
In a category of
Figure 33041DEST_PATH_IMAGE023
The suitability of the individual target color points.
Further, the step of calculating the suitability degree of each color point of the initial class according to the suitability degrees of all the target color points comprises:
obtaining the appropriate degrees of all target color points according to the color difference value sequence of all target color points in each initial class; fitting all target color points and the suitability degrees of all the target color points by a least square method to obtain a suitability degree prediction formula, wherein the suitability degree prediction formula is a ternary cubic polynomial; and obtaining the appropriateness of all the color points of each initial category through an appropriateness prediction formula according to the brightness values and the color values of the color points.
Further, the step of clustering all pixel points in the ferrograph image according to all color points corresponding to the final category to obtain segmentation results of all pixel points in the ferrograph image includes:
obtaining each color point contained in each final category, and obtaining all pixel points corresponding to each color point in the ferrographic image, namely obtaining all pixel points corresponding to all color points contained in each final category; and recording a set formed by all pixel points corresponding to a final category as a category of the ferrographic image to obtain all categories of the ferrographic image, wherein the categories are used as segmentation results of all pixel points in the ferrographic image.
The embodiment of the invention at least has the following beneficial effects:
by changing a category division method of a k-means algorithm in clustering, namely performing category division based on a minimum color difference value, a clustering result obtained by each iteration is more consistent with human vision; and when a new clustering center is obtained before each iteration, based on the appropriate degree of the color points, a new clustering center is obtained by a screening and fitting method, so that the iteration is performed, the final segmentation effect can better accord with the expected segmentation effect of human eye vision, the segmentation effect of the iron spectrum image of the abrasive dust based on color segmentation is improved, and a better basis is provided for the wear condition monitoring and fault diagnosis of equipment.
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 flowchart illustrating steps of a method for intelligently segmenting a ferrographic image based on machine vision according to an embodiment of the present invention.
Detailed Description
To further illustrate the technical means and effects of the present invention adopted to achieve the predetermined objects, the following detailed description of the embodiments, structures, features and effects of the intelligent segmentation method for ferrographic images based on machine vision according to the present invention will be provided with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" refers to not necessarily the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The specific scheme of the intelligent segmentation method for the ferrographic image based on the machine vision provided by the invention is specifically described below by combining the attached drawings.
Referring to fig. 1, a flowchart illustrating steps of a method for intelligent segmentation of ferrographic images based on machine vision according to an embodiment of the present invention is shown, where the method includes the following steps:
and S001, obtaining an iron spectrum image through a microscopic module, and converting all pixel points in the iron spectrum image into color points in a Lab color space.
It should be noted that the obtained ferrographic image is used for subsequent analysis, and since the color distribution in the Lab color space better conforms to the visual distribution of human eyes, the ferrographic image is converted from the RGB color space to the Lab color space, so that a better color clustering result is obtained based on the converted ferrographic image in the subsequent process.
(1) And obtaining a ferrographic image through a microscopic module.
The specific method comprises the following steps: the method comprises the steps of collecting an oil sample in equipment, placing the oil sample on a microscopic observation platform, adopting a magnification of 100 times in the scheme, and obtaining an iron spectrum image of abrasive dust in the oil sample by a microscopic method, wherein an implementer can adjust the magnification of the microscopic observation platform according to a specific implementation scene.
(2) And converting all pixel points in the ferrogram image into color points in the Lab color space.
The specific method comprises the following steps: converting each pixel point in the ferrogram image from RGB color space to XYZ color space, and then converting from XYZ color space to Lab color space to obtain the brightness value of each pixel point in the ferrogram image
Figure DEST_PATH_IMAGE027
Sum color value
Figure 405116DEST_PATH_IMAGE028
And then converting all pixel points in the ferrograph image into color points in the Lab color space, wherein a plurality of pixel points in the ferrograph image correspond to one color point in the Lab color space, and conversely, one color point in the Lab color space corresponds to a plurality of pixel points in the ferrograph image.
And S002, obtaining all initial clustering centers, calculating a first color difference value according to the tolerance of the initial clustering centers, and obtaining all initial categories of all color points according to all the initial clustering centers and the first color difference value.
It should be noted that, because the existing method for clustering the ferrographic image based on color to realize color segmentation of the ferrographic image, pixel points in the ferrographic image that should belong to the same category are classified into different categories, which results in a poor color segmentation result of the ferrographic image. The existing clustering method is only based on color difference for clustering, the color difference is not considered and does not represent the difference of human vision, although the color values of the pixel points of the ferrographic image of one grinding dust are different, the color values of the grinding dust during human vision are approximate, so that when color clustering segmentation is carried out, clustering needs to be carried out according to the tolerance of the color points in order to obtain the color segmentation result of the ferrographic image which is more in line with the human vision effect, therefore, the color difference value needs to be calculated according to the tolerance of the color points, and clustering is further carried out based on the color difference value of the color points.
(1) All initial cluster centers are obtained.
From all color points in the Lab color space, a first number of color points, denoted as initial cluster centers, is randomly selected, which in this embodiment is 3.
(2) And acquiring the latitude of the color points.
It should be noted that, in the Lab color space, the variation range of the color difference that is not perceived by human eyes is called color latitude, and therefore, when the difference between the color of one color point and the color of another color point is smaller than the latitude, the difference between the colors of the two color points is not perceived by human eyes, and therefore, when clustering ferrographic images based on colors, the color points having the color difference smaller than the latitude can be clustered together.
In this embodiment, the latitude of the color point is obtained according to the macadam circle of the color point, which specifically includes: and acquiring the coordinates of the color points on the chromaticity diagram, acquiring corresponding areas of the color points on the chromaticity diagram according to the coordinates of the color points and MacAdam circles of the color points, and recording the areas as the tolerance of the color points.
(3) And calculating a first chromatic aberration value according to the tolerance of the initial clustering center.
It should be noted that, in consideration of the recognition of the color difference by the human vision, the scheme obtains the color difference value calculation formula of the color point according to the latitude of the color point in order to obtain the color segmentation result of the iron spectrum image which is expected by the human vision because the color difference has a certain color tolerance range.
And recording the color difference value between the initial clustering center and the color point in the Lab color space as a first color difference value, wherein the calculation method of the first color difference value comprises the following steps: obtaining the tolerance of the initial clustering center, judging whether the color points in the Lab color space are within the tolerance of the initial clustering center, and if so, determining that the first hue difference value is
Figure 6999DEST_PATH_IMAGE001
(ii) a Otherwise, the first color difference value is:
Figure 93029DEST_PATH_IMAGE003
wherein the content of the first and second substances,
Figure 463967DEST_PATH_IMAGE004
is the luminance value of the initial cluster center,
Figure 475786DEST_PATH_IMAGE005
is the brightness value of a color point in the Lab color space,
Figure 564964DEST_PATH_IMAGE006
is the color value of the initial cluster center,
Figure 953220DEST_PATH_IMAGE007
is the color value of a color point in the Lab color space,
Figure 178665DEST_PATH_IMAGE008
is a first hue difference value;
(4) And obtaining all initial categories of all color points according to all initial cluster centers and the first color difference value.
Calculating the color difference value of each initial clustering center and each color point in the Lab color space; according to the first color difference value of the color point and each initial clustering center, based on the minimum color difference value principle of the K-means clustering algorithm, the color point is divided into the initial clustering centers corresponding to the minimum first color difference value, a set formed by all the color points corresponding to each initial clustering center is marked as an initial category, and all the initial categories of all the color points are obtained.
S003, all the target color points of each initial class are obtained, the color difference value sequence of each target color point is calculated, the appropriateness degree of each target color point is obtained according to the color difference value sequence of the target color points, the appropriateness degree of each color point of each initial class is obtained according to the appropriateness degree of each target color point of each initial class, and then a new clustering center of each initial class is obtained.
(1) All target color points of each initial class are obtained.
Randomly extracting a second number of color points from all the color points of each initial class, and taking the sampled color points as target color points of the initial class, wherein the second number is the product of the number of all the color points of each initial class and a sampling percentage, and in the embodiment, the sampling percentage is
Figure DEST_PATH_IMAGE029
。
(2) A sequence of color difference values for each target color point is calculated.
In this embodiment, the color difference value between the target color point and the other color points in the initial category is recorded as a second color difference value, and the calculation method of the second color difference value is as follows: obtaining the tolerance of the target color point, judging whether other color points in the initial category are within the tolerance of the target color point, and if the other color points in the initial category are within the tolerance of the target color point, determining that the second color difference value is
Figure 892543DEST_PATH_IMAGE009
(ii) a Otherwise the second hue value is:
Figure 233133DEST_PATH_IMAGE011
wherein the content of the first and second substances,
Figure 159500DEST_PATH_IMAGE012
for the luminance value of the target color point,
Figure 239452DEST_PATH_IMAGE013
the luminance values of the other color points in the initial class,
Figure 858652DEST_PATH_IMAGE014
is the color value of the target color point,
Figure 656844DEST_PATH_IMAGE014
color values for other color points in the initial category,
Figure 652482DEST_PATH_IMAGE015
is the second hue value.
And arranging all second color difference values of the target color point and all other color points in the initial category from small to large, and recording the formed sequence as a color difference value sequence of the target color point.
(3) And obtaining the appropriate degree of the target color point according to the color difference value sequence of the target color point.
It should be noted that, the more stable the sequence of color difference values of the target color point, and the more consistent the color difference values of the target color point and other color points in the initial category, the more consistent the difference of all color points in the initial category in human vision, the more suitable the target color point is as a new clustering center, so that the suitable degree of the target color point is obtained according to the sequence of color difference values of the target color point, and thus a new clustering center is obtained.
In this embodiment, the first step is calculated
Figure 586940DEST_PATH_IMAGE021
In an initial category
Figure 878506DEST_PATH_IMAGE023
The suitability of the target color point is determined by
Figure 163994DEST_PATH_IMAGE021
In an initial category
Figure 432164DEST_PATH_IMAGE023
The color difference value sequence of each target color point is obtained, and the specific calculation formula is as follows:
Figure 752287DEST_PATH_IMAGE017
wherein the content of the first and second substances,
Figure 713290DEST_PATH_IMAGE018
an exponential function with a natural constant as the base is represented,
Figure 220494DEST_PATH_IMAGE019
is as follows
Figure 292356DEST_PATH_IMAGE021
In a category of
Figure 242818DEST_PATH_IMAGE023
The variance of all color difference values in the sequence of color difference values for each target color point,
Figure 109143DEST_PATH_IMAGE024
is as follows
Figure 369223DEST_PATH_IMAGE021
In a category of
Figure 979196DEST_PATH_IMAGE023
The difference of the maximum value and the minimum value of all color difference values in the sequence of color difference values of the target color point,
Figure 8332DEST_PATH_IMAGE025
is as follows
Figure 311137DEST_PATH_IMAGE021
In a category of
Figure 58513DEST_PATH_IMAGE023
The mean of all color difference values in the sequence of color difference values for each target color point,
Figure 973642DEST_PATH_IMAGE026
is as follows
Figure 857284DEST_PATH_IMAGE021
In a category of
Figure 330991DEST_PATH_IMAGE023
The suitability of the individual target color points.
Wherein the content of the first and second substances,
Figure 300084DEST_PATH_IMAGE019
the smaller the sequence of color difference values indicating the target color point is more stable, the more appropriate the target color point is
Figure 517439DEST_PATH_IMAGE026
The larger the target color point is, the more suitable the target color point is as a new clustering center;
Figure 521167DEST_PATH_IMAGE024
smaller values indicate more stable color difference value sequence of target color pointsAnd, if so, the appropriate degree of the target color point
Figure 900195DEST_PATH_IMAGE026
The larger the target color point is, the more suitable the target color point is as a new clustering center;
Figure 356585DEST_PATH_IMAGE025
the smaller the sequence of color difference values indicating the more consistent the target color point, the more appropriate the target color point is
Figure 610586DEST_PATH_IMAGE026
The larger the target color point, the more suitable it is as a new cluster center.
(4) And obtaining the appropriateness of all the color points in each initial class according to the appropriateness of all the target color points in each initial class, and further obtaining a new clustering center of each initial class.
In the embodiment, the appropriate degrees of all the target color points are obtained according to the color difference value sequence of all the target color points in each initial category; fitting all target color points and the suitable degrees of all the target color points by a least square method to obtain a suitable degree prediction formula; according to a fitness degree prediction formula, the fitness degree prediction formula is a ternary cubic polynomial; and obtaining the appropriateness of all the color points of each initial category through an appropriateness prediction formula according to the brightness values and the color values of the color points, and marking the color point with the maximum appropriateness in all the color points of each initial category as a new cluster center.
And S004, obtaining the final classification of all color points.
In this embodiment, the new cluster center is used as the initial cluster center, and S002 and S003 are repeatedly executed until the initial cluster center is not changed any more, and the obtained initial category is used as the final category.
And S005, clustering all the pixel points in the ferrograph image according to the final category to obtain the segmentation result of all the pixel points in the ferrograph image.
Obtaining each color point contained in each final category, and obtaining all pixel points corresponding to the color points in the ferrographic image, namely obtaining all pixel points corresponding to all the color points contained in each final category; and recording a set formed by all pixel points corresponding to a final category as a category of the ferrographic image to obtain all categories of the ferrographic image, wherein the categories are used as segmentation results of all the pixel points in the ferrographic image, and the segmentation results of the ferrographic image are more in line with the segmentation results expected by human vision.
In summary, the present invention converts all pixel points in the ferrographic image into color points in the Lab color space; obtaining all initial categories of all color points according to the initial clustering center and the first color difference value; obtaining the appropriateness of each target color point in each initial category, obtaining the appropriateness of all color points of each initial category according to an appropriateness prediction formula, and further obtaining a new clustering center; obtaining the final categories of all color points through multiple iterations; and obtaining a segmentation result of the ferrographic image according to the final classification of all the color points. The method of the invention enables the segmentation result of the ferrographic image to better accord with the expected segmentation effect of human vision, and provides better basis for the wear condition monitoring and fault diagnosis of equipment.
It should be noted that: the precedence order of the above embodiments of the present invention is only for description, and does not represent the merits of the embodiments. And specific embodiments thereof have been described above. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments.
The above embodiments are only used to illustrate the technical solutions of the present application, and not to limit the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications or substitutions do not cause the essential features of the corresponding technical solutions to depart from the scope of the technical solutions of the embodiments of the present application, and are intended to be included within the scope of the present application.

Claims (6)

1. The intelligent segmentation method of the ferrographic image based on the machine vision is characterized by comprising the following steps:
obtaining an iron spectrum image of abrasive dust in the oil sample through a microscopic module; converting all pixel points in the ferrographic image into color points in a Lab color space;
randomly selecting a first number of color points from all color points in the Lab color space, and recording the color points as initial clustering centers;
s1: obtaining all initial classes of all color points according to all initial cluster centers, including:
calculating all first color difference values of all initial clustering centers and all color points in the Lab color space; clustering all the color points by using a K-means clustering algorithm according to all the first color difference values and the initial clustering centers to obtain all initial categories of all the color points;
s2, obtaining all new clustering centers according to all initial categories, including:
for any one initial class, randomly sampling a plurality of color points from all the color points of the initial class, and marking as target color points; calculating all second color difference values of all target color points and all other color points in the initial category, and obtaining the appropriate degree of the target color points according to all the second color difference values; calculating the appropriateness of each color point of the initial class according to the appropriateness of all the target color points; recording the color point with the maximum suitable degree in all the color points of the initial category as a new clustering center; for all initial categories, obtaining all new clustering centers;
repeatedly executing S1 and S2 by taking the new clustering center as an initial clustering center until the initial clustering center is not changed any more, and taking the obtained initial category as a final category;
and clustering all pixel points in the ferrograph image according to all color points corresponding to the final category to obtain the segmentation result of all pixel points in the ferrograph image.
2. The intelligent machine-vision-based ferrographic image segmentation method according to claim 1, wherein the first color difference value and the second color difference value comprise:
obtaining the tolerance of the initial clustering center, judging whether the color point in the Lab color space is within the tolerance of the initial clustering center or not, and if the color point in the Lab color space is within the tolerance of the initial clustering center, determining that the first hue difference value is
Figure DEST_PATH_IMAGE002
(ii) a Otherwise, the first color difference value is:
Figure DEST_PATH_IMAGE004
wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE006
is the luminance value of the initial cluster center,
Figure DEST_PATH_IMAGE008
is the brightness value of a color point in the Lab color space,
Figure DEST_PATH_IMAGE010
is the color value of the initial cluster center,
Figure DEST_PATH_IMAGE012
color values for color points in Lab color space，
Figure DEST_PATH_IMAGE014
Is a first hue difference value;
obtaining the tolerance of the target color point, judging whether other color points in the initial category are within the tolerance of the target color point, and if the other color points in the initial category are within the tolerance of the target color point, determining that the second color difference value is
Figure DEST_PATH_IMAGE016
(ii) a Otherwise the second hue value is:
Figure DEST_PATH_IMAGE018
wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE020
for the luminance value of the target color point,
Figure DEST_PATH_IMAGE022
the luminance values of the other color points in the initial class,
Figure DEST_PATH_IMAGE024
is the color value of the target color point,
Figure 433588DEST_PATH_IMAGE024
the color values of the other color points in the initial category,
Figure DEST_PATH_IMAGE026
is the second hue value.
3. The machine-vision-based intelligent segmentation method for ferrographic images as claimed in claim 1, wherein the step of clustering all color points by using a K-means clustering algorithm according to all the first color difference values and the initial clustering centers to obtain all the initial classes of all the color points comprises:
according to the first color difference value of the color point and each initial clustering center, based on the minimum color difference value principle of the K-means clustering algorithm, the color point is divided into the initial clustering centers corresponding to the minimum first color difference value, a set formed by all the color points corresponding to each initial clustering center is marked as an initial category, and all the initial categories of all the color points are obtained.
4. The intelligent segmentation method for ferrographic images based on machine vision according to claim 1, wherein the step of obtaining the suitability of the target color point according to all the second color difference values comprises:
recording a sequence of all second color difference values of the target color point and all other color points in the initial category as a color difference value sequence of the target color point, and obtaining the appropriateness degree of the target color point according to the color difference value sequence of the target color point, wherein the calculation formula of the appropriateness degree of the target color point is as follows:
Figure DEST_PATH_IMAGE028
wherein, the first and the second end of the pipe are connected with each other,
Figure DEST_PATH_IMAGE030
an exponential function with a natural constant as the base is represented,
Figure DEST_PATH_IMAGE032
is as follows
Figure DEST_PATH_IMAGE034
In a category of
Figure DEST_PATH_IMAGE036
The variance of all color difference values in the sequence of color difference values for each target color point,
Figure DEST_PATH_IMAGE038
is as follows
Figure 826261DEST_PATH_IMAGE034
In a category of
Figure 35526DEST_PATH_IMAGE036
The difference between the maximum value and the minimum value of all color difference values in the sequence of color difference values for each target color point,
Figure DEST_PATH_IMAGE040
is as follows
Figure 860262DEST_PATH_IMAGE034
In a category of
Figure 838582DEST_PATH_IMAGE036
The mean of all color difference values in the sequence of color difference values for each target color point,
Figure DEST_PATH_IMAGE042
is as follows
Figure 356151DEST_PATH_IMAGE034
In a category of
Figure 551247DEST_PATH_IMAGE036
The suitability of the individual target color points.
5. The machine-vision-based intelligent segmentation method for ferrographic images as claimed in claim 1, wherein the step of calculating the suitability degree of each color point of the initial class according to the suitability degrees of all target color points comprises:
obtaining the appropriate degrees of all target color points according to the color difference value sequence of all target color points in each initial class; fitting all target color points and the suitability degrees of all the target color points by a least square method to obtain a suitability degree prediction formula, wherein the suitability degree prediction formula is a ternary cubic polynomial; and obtaining the appropriateness of all the color points of each initial category through an appropriateness prediction formula according to the brightness values and the color values of the color points.
6. The intelligent segmentation method for a ferrographic image based on machine vision according to claim 1, wherein the step of clustering all pixel points in the ferrographic image according to all color points corresponding to the final classification to obtain the segmentation result of all pixel points in the ferrographic image comprises:
obtaining each color point contained in each final category, and obtaining all pixel points corresponding to each color point in the ferrographic image, namely obtaining all pixel points corresponding to all color points contained in each final category; and recording a set formed by all pixel points corresponding to a final category as a category of the ferrographic image to obtain all categories of the ferrographic image, wherein the categories are used as segmentation results of all pixel points in the ferrographic image.
CN202211241929.4A 2022-10-11 2022-10-11 Intelligent segmentation method for ferrographic image based on machine vision Active CN115311276B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211241929.4A CN115311276B (en) 2022-10-11 2022-10-11 Intelligent segmentation method for ferrographic image based on machine vision

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211241929.4A CN115311276B (en) 2022-10-11 2022-10-11 Intelligent segmentation method for ferrographic image based on machine vision

Publications (2)

Publication Number Publication Date
CN115311276A true CN115311276A (en) 2022-11-08
CN115311276B CN115311276B (en) 2023-01-17

Family

ID=83868114

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211241929.4A Active CN115311276B (en) 2022-10-11 2022-10-11 Intelligent segmentation method for ferrographic image based on machine vision

Country Status (1)

Country Link
CN (1) CN115311276B (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020102017A1 (en) * 2000-11-22 2002-08-01 Kim Sang-Kyun Method and apparatus for sectioning image into plurality of regions
CN102360494A (en) * 2011-10-18 2012-02-22 中国科学院自动化研究所 Interactive image segmentation method for multiple foreground targets
CN110910417A (en) * 2019-10-29 2020-03-24 西北工业大学 Weak and small moving target detection method based on super-pixel adjacent frame feature comparison

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020102017A1 (en) * 2000-11-22 2002-08-01 Kim Sang-Kyun Method and apparatus for sectioning image into plurality of regions
CN102360494A (en) * 2011-10-18 2012-02-22 中国科学院自动化研究所 Interactive image segmentation method for multiple foreground targets
CN110910417A (en) * 2019-10-29 2020-03-24 西北工业大学 Weak and small moving target detection method based on super-pixel adjacent frame feature comparison

Also Published As

Publication number Publication date
CN115311276B (en) 2023-01-17

Similar Documents

Publication Publication Date Title
US7528991B2 (en) Method of generating a mask image of membership of single pixels to certain chromaticity classes and of adaptive improvement of a color image
JP2008546990A (en) How to split white blood cells
WO2004027695A1 (en) Color naming, color categorization and describing color composition of images
Hervé et al. Statistical color texture descriptors for histological images analysis
CA2851498C (en) System and method for digital image signal compression using intrinsic images
CN110689587A (en) Underwater image enhancement method based on color correction and detail enhancement
CN114648594B (en) Textile color detection method and system based on image recognition
CN107180439B (en) Color cast characteristic extraction and color cast detection method based on Lab chromaticity space
CN114494256A (en) Electric wire production defect detection method based on image processing
Jiang et al. Color image enhancement with brightness preservation using a histogram specification approach
Kapoor et al. Capturing banding in images: Database construction and objective assessment
CN111079637A (en) Method, device and equipment for segmenting rape flowers in field image and storage medium
CN115311276B (en) Intelligent segmentation method for ferrographic image based on machine vision
CN106415596B (en) image conversion based on segmentation
US20080247647A1 (en) Systems and methods for segmenting an image based on perceptual information
Hu et al. No reference quality assessment for Thangka color image based on superpixel
CN103927544A (en) Machine vision grading method for ginned cotton rolling quality
Khediri et al. Comparison of image segmentation using different color spaces
CN114972849A (en) Glioma type identification method, model training method, device and equipment
CN113902013A (en) Hyperspectral classification method based on three-dimensional convolutional neural network and superpixel segmentation
Lezoray et al. Automatic segmentation and classification of cells from broncho alveolar lavage
Kandil et al. Automatic segmentation of acute leukemia cells
Haridas et al. Pap smear image enhancement using diffusion stop function based CLAHE algorithm
Guo et al. A Novel Low-light Image Enhancement Algorithm Based On Information Assistance
Fang et al. Image segmentation using adaptively selected color space

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