CN114862854B - Ship electrical accessory defect detection method - Google Patents

Ship electrical accessory defect detection method Download PDF

Info

Publication number
CN114862854B
CN114862854B CN202210791378.2A CN202210791378A CN114862854B CN 114862854 B CN114862854 B CN 114862854B CN 202210791378 A CN202210791378 A CN 202210791378A CN 114862854 B CN114862854 B CN 114862854B
Authority
CN
China
Prior art keywords
pixel
pixel point
bump
bulge
mean value
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202210791378.2A
Other languages
Chinese (zh)
Other versions
CN114862854A (en
Inventor
房素平
董凤凤
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.)
SHANGHAI QUNLE SHIP ACCESSORIES QIDONG CO Ltd
Original Assignee
SHANGHAI QUNLE SHIP ACCESSORIES QIDONG 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 SHANGHAI QUNLE SHIP ACCESSORIES QIDONG CO Ltd filed Critical SHANGHAI QUNLE SHIP ACCESSORIES QIDONG CO Ltd
Priority to CN202210791378.2A priority Critical patent/CN114862854B/en
Publication of CN114862854A publication Critical patent/CN114862854A/en
Application granted granted Critical
Publication of CN114862854B publication Critical patent/CN114862854B/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
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
    • 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
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/761—Proximity, similarity or dissimilarity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30108—Industrial image inspection
    • G06T2207/30164—Workpiece; Machine component

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Computing Systems (AREA)
  • Multimedia (AREA)
  • Software Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Probability & Statistics with Applications (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Quality & Reliability (AREA)
  • Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
  • Image Analysis (AREA)

Abstract

The invention relates to the field of artificial intelligence, and provides a method for detecting defects of ship electrical equipment accessories, which adopts a pattern recognition method to recognize the defects and comprises the following steps: acquiring brightness images of the surface of the cabinet body of the electric appliance control cabinet at different angles; obtaining a brightness distribution vector of each position pixel point; obtaining the similarity mean value of the pixel points at each position; obtaining the category of each pixel point; obtaining a first similarity mean value; obtaining a second similarity mean value; obtaining the reliability; calculating the bulge probability of each pixel point; and judging whether the pixel points are bulge pixels. The invention improves the detection precision of bulge defects.

Description

Ship electrical accessory defect detection method
Technical Field
The invention relates to the field of artificial intelligence, in particular to a method for detecting defects of ship electrical fittings.
Background
With the development of social economy and the improvement of the global trade and business transaction amount, the use amount of ships is improved. With the development of the technology and the improvement of the automation of the ship, the use amount of electric equipment of the ship is increased. Because the ship sails in seawater and is in a high-humidity and high-salt environment for a long time, if the power equipment is not well protected, the ship is easy to rust and damage, and the service life is shortened. The electrical control cabinet is an important protection device of electrical equipment, and can play the best protection role only when the electrical control cabinet is not damaged. Meanwhile, the paint surface of the electric control cabinet is an important barrier for blocking a high-humidity environment, and when the electric control cabinet has defects, the paint surface is easy to fall off, so that the protection of the electric control cabinet on electric equipment is influenced, and the paint surface condition of the electric control cabinet needs to be paid attention to. Paint surface bulging is a common paint surface defect that has a low contrast ratio and an insignificant edge line of demarcation. Meanwhile, the production workplaces of the electric control cabinet are mostly in natural light, the surface of the electric control cabinet is smooth and is sensitive to illumination, and therefore, the bulge area of the paint surface of the electric control cabinet is difficult to accurately divide.
The traditional defect detection method mostly adopts edge detection and threshold segmentation, wherein when the edge detection method is used for processing the bulge defects under the natural light, part of boundaries, particularly bright-face boundaries of the bulge, are difficult to position due to the interference of illumination, and the plane area of the cabinet body has large gray scale change due to complex illumination, so that false edges caused by light change can be detected by using the edge detection, and the bulge area is difficult to accurately position by using the edge detection. For the threshold segmentation method, since light rays cause complex gray scale changes and the bulge region also has complex gray scale changes, it is difficult to segment an accurate bulge region by simple threshold segmentation. In view of the above, the method provided by the invention considers the complex illumination influence, adopts a pattern recognition method to identify the defects, deals with the light influence through the multi-view image, obtains the probability that each pixel belongs to the bulge by using the multi-view image, and determines whether each pixel point is a bulge pixel or not by using the bulge probability, thereby accurately positioning the bulge defect area.
Disclosure of Invention
The invention provides a defect detection method for ship electrical fittings, which aims to solve the problem that the bulge defect area is difficult to accurately detect in the prior art.
The invention discloses a method for detecting defects of ship electrical fittings, which adopts the following technical scheme:
acquiring brightness images of the surface of the cabinet body of the electric appliance control cabinet at different angles;
obtaining a brightness distribution vector of each position pixel point through each order moment of each position pixel point at different angle brightness values;
obtaining the similarity mean value of the pixel points at each position through the brightness distribution vectors of the pixel points at each position and the pixel points at other positions;
clustering the pixel points of all positions by using the similarity mean value of the pixel points of each position to obtain the category to which each pixel point belongs, wherein the categories include: suspected bump pixel clustering and suspected non-bump pixel clustering;
obtaining a first similarity mean value according to the brightness distribution vector of each pixel point and the brightness distribution vector of each pixel point in the suspected bump pixel cluster;
obtaining a second similarity mean value according to the brightness distribution vector of each pixel point and the brightness distribution vector of each pixel point in the suspected non-bump pixel cluster;
obtaining the reliability by using the difference value of the first similarity mean value and the second similarity mean value obtained by each pixel point and the similarity mean value variance of all position pixel points;
calculating the bulge probability of each pixel point according to the first similarity mean value, the second similarity mean value and the reliability obtained by each pixel point and the category to which the pixel point belongs;
and judging whether each pixel point is a bulge pixel or not according to the bulge probability value of each pixel point and the bulge probability threshold value.
Further, in the method for detecting the defects of the ship electrical fittings, the method for obtaining the similarity mean value of the pixel points at each position comprises the following steps:
calculating the cosine similarity of the brightness distribution vectors of the pixel points at the positions and the pixel points at other positions according to the brightness distribution vectors of the pixel points at the positions and the pixel points at other positions;
and obtaining the similarity mean value of the pixel points at each position according to the cosine similarity of the brightness distribution vectors of the pixel points at each position and the pixel points at other positions.
Further, the method for detecting the defects of the ship electrical fittings comprises the following steps:
obtaining a similarity difference value of each pixel point according to the first similarity mean value and the second similarity mean value obtained by each pixel point and the pixel point;
and obtaining the reliability by using the similarity mean variance of all the position pixel points and the similarity difference of all the pixel points.
Further, in the method for detecting the defects of the ship electrical fittings, the expression of the bulge probability of each pixel point is as follows:
the expression of the bump probability of a suspected bump pixel is:
Figure 65329DEST_PATH_IMAGE002
in the formula:
Figure 100002_DEST_PATH_IMAGE003
indicating the first of a cluster of suspected bump pixels
Figure 100002_DEST_PATH_IMAGE005
The similarity mean value of each pixel point and the similar pixels,
Figure 867456DEST_PATH_IMAGE006
indicating the first of a cluster of suspected bump pixels
Figure 100002_DEST_PATH_IMAGE007
The similarity mean value of each pixel point and non-similar pixels,
Figure 466934DEST_PATH_IMAGE008
the degree of confidence is expressed in the form of,
Figure DEST_PATH_IMAGE009
representing the similarity mean variance of all the position pixel points,
Figure 974226DEST_PATH_IMAGE010
representing the first in a suspected bump pixel cluster
Figure 624650DEST_PATH_IMAGE007
The bump probability of each pixel point;
the expression for the bump probability for a suspected non-bump pixel is:
Figure 532432DEST_PATH_IMAGE012
in the formula:
Figure 100002_DEST_PATH_IMAGE013
indicating a suspected non-bump pixel cluster
Figure DEST_PATH_IMAGE015
The similarity mean value of each pixel point and the similar pixels,
Figure 838780DEST_PATH_IMAGE016
indicating a suspected non-bump pixel cluster
Figure 123000DEST_PATH_IMAGE015
The similarity mean value of each pixel point and non-similar pixels,
Figure DEST_PATH_IMAGE017
indicating a suspected non-bump pixel cluster
Figure 565613DEST_PATH_IMAGE015
And (4) the bulge probability of each pixel point.
Further, the method for detecting the defects of the ship electrical fittings, which judges whether each pixel point is a bulge pixel according to the bulge probability value of each pixel point and the bulge probability threshold value, comprises the following steps:
if the bulge probability value of the pixel point is larger than the bulge probability threshold value, judging the pixel point as a bulge pixel;
and if the swelling probability value of the pixel point is less than or equal to the swelling probability threshold value, judging that the pixel point is a non-swelling pixel.
Further, in the method for detecting the defects of the ship electrical accessories, the suspected bump pixel clustering is a clustering with a small number of pixel points obtained after clustering the pixel points at all positions; the suspected non-bulge pixel clustering is a clustering with a large number of pixel points obtained after clustering the pixel points at all positions.
The beneficial effects of the invention are: the method adopts a pattern recognition method to recognize the defects, firstly obtains the probability that each pixel point belongs to the bulge by using the multi-view image, and then determines whether each pixel point is a bulge pixel or not by using the probability of the bulge, thereby accurately positioning the bulge defect area, simultaneously avoiding the interference of complex illumination on the detection result, improving the precision of bulge defect detection and accurately positioning the bulge defect area compared with the prior art.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a schematic flow chart illustrating an embodiment of a method for detecting defects of an electrical accessory of a ship according to the present invention;
fig. 2 is a schematic view of a cabinet surface picture.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Example 1
An embodiment of a method for detecting a defect of an electrical accessory of a ship according to the present invention is shown in fig. 1, and includes:
the present invention is directed to the following scenarios: the method comprises the steps of firstly arranging a camera above a cabinet body, collecting pictures on the surface of the cabinet body through multiple angles, obtaining the bulge probability of each pixel through analyzing the pictures of the multiple angles, and then determining whether each pixel point is a bulge pixel or not by utilizing the bulge probability.
As shown in fig. 2, due to the influence of light, there is a complex gray scale change on the surface of the cabinet, and at the same time, there is a change in the bright and dark surfaces in the bulge region, and the boundaries of the bulge defects are not clear, and particularly, it is difficult to distinguish the accurate region from the bulge with a small bulge.
101. Acquiring brightness images of the surface of the cabinet body of the electric appliance control cabinet at different angles;
in the embodiment, the accurate region of the bulge on the surface of the cabinet body needs to be detected, the picture of the surface of the cabinet body needs to be collected firstly, and the position of the bulge defect is located through the picture of the surface of the cabinet body.
Because under natural illumination, light distribution is comparatively complicated to cabinet body surface is comparatively smooth, and it is comparatively sensitive to the light, and according to the difficult accurate location of a single picture bulge defect position, therefore need the multi-angle to gather the picture, confirm accurate defect position through many pictures together. A camera is arranged above the surface of the cabinet body, one cabinet body picture is collected at intervals of 1 degree, and 360 pictures are obtained by rotating for one circle.
And then, carrying out color space conversion on the acquired pictures, converting each image from an RGB color space to an HSV color space, and acquiring the image of a V channel (namely a brightness channel).
And determining a perspective transformation matrix according to the relative position of the camera and the surface of the cabinet body, and eliminating the perspective deformation through inverse transformation.
Through the steps, cabinet surface brightness images at different angles are obtained.
102. Obtaining a brightness distribution vector of each position pixel point through each order moment of each position pixel point at different angle brightness values;
firstly, a same-position pixel set is obtained, and the specific method comprises the following steps: in 360 multi-angle pictures, extracting the brightness values of the same coordinate position to form a pixel set at the same position.
If the cabinet surface brightness image exists
Figure 644297DEST_PATH_IMAGE018
Three pixels, will
Figure DEST_PATH_IMAGE019
The pixel values (i.e. brightness values) of the pixels in 360 pictures are extracted and composed
Figure 969099DEST_PATH_IMAGE019
Luminance collection of pixels at different angles
Figure 807742DEST_PATH_IMAGE020
In the same way, obtain
Figure DEST_PATH_IMAGE021
Luminance collection of pixels at different angles
Figure 171375DEST_PATH_IMAGE022
And
Figure DEST_PATH_IMAGE023
luminance collection of pixels at different angles
Figure 640534DEST_PATH_IMAGE024
。
Then, the brightness distribution of the pixel points at each position is obtained, and the specific method is as follows: and acquiring an n-dimensional brightness distribution vector of the pixel points at each position, which is formed by the first moment, the second moment, … and the n-order moment of the pixel value of the pixel point set at each position.
For example, by
Figure 967479DEST_PATH_IMAGE019
Luminance collection of pixels at different angles
Figure 609813DEST_PATH_IMAGE020
The corresponding first moment, second moment, …, n-order moment are obtained from the brightness value in (1), and the corresponding first moment, second moment, …, n-order moment are obtained
Figure 89336DEST_PATH_IMAGE019
An n-dimensional luminance distribution vector of pixels.
103. Obtaining the similarity mean value of the pixel points at each position through the brightness distribution vectors of the pixel points at each position and the pixel points at other positions;
in determining the bump probability, it is derived on an empirical assumption that: under the same light line condition, the pixel brightness values of the plane area present similar distribution rules under all angles, and the pixel brightness values of the bulge area present the distribution rules under all angles different from the distribution rules presented by the pixel brightness values of the plane area. Based on the above, the bump probability of each pixel is calculated by analyzing the pixel distribution similarity difference at each position.
Calculating the similarity of the cosine and the sine of the two position pixel points through the brightness distribution vectors of every two position pixel points
Figure DEST_PATH_IMAGE025
And averaging the similarity of the positions and the cosine or the sine of other positions to obtain the similarity average of the pixel points at each position.
Such as calculation
Figure 244242DEST_PATH_IMAGE019
The similarity of the pixel and the residual myth of the pixel in all other positions is calculated, the calculated similarity of the residual myth is averaged, and the average value is the similarity of the residual myth and the pixel in all other positions
Figure 809216DEST_PATH_IMAGE019
The mean value of the similarity of the pixels.
104. Clustering the pixel points of all positions by using the similarity mean value of the pixel points of each position to obtain the category to which each pixel point belongs, wherein the categories include: suspected bump pixel clustering and suspected non-bump pixel clustering;
because the bulge pixel regions and the non-bulge pixel regions are distributed differently, and meanwhile, under normal conditions, the number of non-bulge pixels is large, the similarity mean value of the non-bulge pixels is large, and therefore, the non-bulge pixels are simply distinguished firstly through cluster analysis.
And performing k-means clustering on the similarity mean value of the pixel points at each position, wherein the category number is two. The category with a large number of pixel points is suspected non-bump pixel clustering, and the cluster with a small number of pixel points is suspected bump pixel clustering.
105. Obtaining a first similarity mean value according to the brightness distribution vector of each pixel point and the brightness distribution vector of each pixel point in the suspected bump pixel cluster;
106. obtaining a second similarity mean value according to the brightness distribution vector of each pixel point and the brightness distribution vector of each pixel point in the suspected non-bump pixel cluster;
because the distribution similarity between the suspected bump pixels is large, if the similarity between the pixel point and the similar pixel is large, the possibility that the pixel point is a bump is large, and meanwhile, the distribution similarity between the suspected bump pixel and the non-bump pixel is small, so that the difference of the distribution similarity between the pixel and the non-bump pixel is large, the probability that the pixel point is a bump is large, and the probability of the bump is analyzed based on the probability.
And calculating the similarity mean value of each pixel point and the same type of pixels. The greater the similarity with the similar pixel, the greater the possibility that the pixel point belongs to the similar pixel.
If the target pixel point is a suspected non-bulge pixel, the method for obtaining the similarity mean value comprises the following steps: and calculating cosine similarity of the suspected non-bulge pixels and other suspected non-bulge pixels, and averaging all cosine similarity to obtain a similarity mean value of the suspected non-bulge pixels and similar pixels.
And calculating the similarity mean value of each pixel and the non-homogeneous pixels. The smaller the similarity with the non-homogeneous pixels, the more likely it is that the pixels of this type belong to this category.
If the target pixel point is a suspected non-bump pixel, the method for obtaining the similarity mean value comprises the following steps: and calculating the cosine similarity between the suspected non-bulge pixels and each suspected bulge pixel, and averaging all the cosine similarities to obtain the similarity mean value of the suspected non-bulge pixels and non-similar pixels.
107. Obtaining the reliability by using the difference value of the first similarity mean value and the second similarity mean value obtained by each pixel point and the similarity mean value variance of all position pixel points;
obtaining the variance of the similarity mean of all position pixel points
Figure 989661DEST_PATH_IMAGE009
。
The similarity mean variance refers to the similarity mean of the pixel points at each position obtained by calculating the cosine similarity of the pixel point at each position and the pixels at all other positions before clustering. Calculating the variance of the similarity mean value of all the position pixel points, namely the variance is
Figure 589270DEST_PATH_IMAGE009
。
And subtracting the similarity mean value of each pixel point and the pixels of the same type from the similarity mean value of each pixel point and the pixels of non-same type to obtain the similarity difference value of each pixel point.
The similarity difference and variance of each pixel point are calculated
Figure 446236DEST_PATH_IMAGE009
And comparing the sizes, and counting the ratio of the number of the pixels larger than the variance to the total pixels. The ratio is the reliability of the previous classification
Figure 498506DEST_PATH_IMAGE008
。
108. Calculating the bulge probability of each pixel point according to the first similarity mean value, the second similarity mean value and the reliability obtained by each pixel point and the category to which the pixel point belongs;
calculating the bulge probability:
the expression of the bump probability of a suspected bump pixel is:
Figure DEST_PATH_IMAGE027
in the formula:
Figure 685905DEST_PATH_IMAGE003
indicating the first of a cluster of suspected bump pixels
Figure 392217DEST_PATH_IMAGE007
The similarity mean value of each pixel point and the similar pixels,
Figure 170817DEST_PATH_IMAGE006
indicating the first of a cluster of suspected bump pixels
Figure 710383DEST_PATH_IMAGE007
The similarity mean value of each pixel point and non-homogeneous pixels,
Figure 685161DEST_PATH_IMAGE008
the confidence level of the early classification is represented,
Figure 790521DEST_PATH_IMAGE009
representing the variance of the mean of similarity of all position pixels,
Figure 943284DEST_PATH_IMAGE010
indicating the first of a cluster of suspected bump pixels
Figure 704567DEST_PATH_IMAGE007
The bump probability of each pixel.
Figure 279774DEST_PATH_IMAGE028
Indicating the first in the category of the bulge
Figure 239639DEST_PATH_IMAGE007
The higher the degree of the pixel according with the early-stage classification result, the higher the degree of the pixel according with the early-stage classification result shows that the probability of the pixel being a bulge is higher, meanwhile, the stronger the similarity of the pixel and the same type of pixel can be reflected by the formula, and the worse the similarity with the non-same type of pixel is, the higher the probability of the pixel belonging to the bulge is,
Figure DEST_PATH_IMAGE029
to take into account the confidence of earlier classification
Figure 32146DEST_PATH_IMAGE007
The probability of a bump of a pixel,
Figure 795572DEST_PATH_IMAGE008
the larger the result of the initial classification, the more accurate the suspected bump class pixel is, the more likely it is that the bump pixel belongs to.
The expression for the bump probability for a suspected non-bump pixel is:
Figure DEST_PATH_IMAGE031
in the formula:
Figure 925202DEST_PATH_IMAGE013
indicating a suspected non-bump pixel cluster
Figure 942836DEST_PATH_IMAGE015
The mean value of the similarity between each pixel point and the similar pixels represents the first of the suspected non-bump pixel clusters
Figure 499720DEST_PATH_IMAGE015
The similarity mean value of each pixel point and non-similar pixels,
Figure 481932DEST_PATH_IMAGE017
indicating a suspected non-bump pixel cluster
Figure 352936DEST_PATH_IMAGE015
The bump probability of each pixel.
Figure 21815DEST_PATH_IMAGE032
Is shown as
Figure 749600DEST_PATH_IMAGE015
The degree to which each suspected non-bump pixel does not meet the initial classification indicates that the pixel has a higher probability of being a bump if the pixel does not meet the initial classification, and the formula also indicates that the pixel has a lower degree of similarity with pixels of the same type, and the pixel has a higher probability of being a bump if the degree of similarity with pixels of the same type is higher.
Figure DEST_PATH_IMAGE033
Taking into account initial classification confidence
Figure 690880DEST_PATH_IMAGE015
Probability that a pixel belongs to a bump pixel, where
Figure 896733DEST_PATH_IMAGE034
The larger the indication, the less trustworthy, i.e. the greater the likelihood that a non-bump class pixel belongs to a bump pixel.
109. And judging whether the pixel point is a bulge pixel or not according to the bulge probability value of each pixel point and the bulge probability threshold value.
Setting a bulge probability threshold, and if the bulge probability value of a pixel point is greater than the bulge probability threshold, judging the pixel point as a bulge pixel;
and if the swelling probability value of the pixel point is less than or equal to the swelling probability threshold value, judging that the pixel point is a non-swelling pixel.
The invention has the beneficial effects that: the method adopts a pattern recognition method to recognize the defects, firstly obtains the probability that each pixel point belongs to the bulge by using the multi-view image, and then determines whether each pixel point is a bulge pixel or not by using the probability of the bulge, thereby accurately positioning the bulge defect area, simultaneously avoiding the interference of complex illumination on the detection result, improving the precision of bulge defect detection and accurately positioning the bulge defect area compared with the prior art.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and should not be taken as limiting the scope of the present invention, which is intended to cover any modifications, equivalents, improvements, etc. within the spirit and scope of the present invention.

Claims (6)

1. A method for detecting defects of ship electrical accessories is characterized by comprising the following steps:
acquiring brightness images of the surface of the cabinet body of the electric appliance control cabinet at different angles;
obtaining a brightness distribution vector of each position pixel point through each order moment of each position pixel point at different angle brightness values;
obtaining the similarity mean value of the pixel points at each position through the brightness distribution vectors of the pixel points at each position and the pixel points at other positions;
clustering the pixel points of all positions by using the similarity mean value of the pixel points of each position to obtain the category to which each pixel point belongs, wherein the categories include: suspected bump pixel clustering and suspected non-bump pixel clustering;
obtaining a first similarity mean value according to the brightness distribution vector of each pixel point and the brightness distribution vector of each pixel point in the suspected bump pixel cluster;
obtaining a second similarity mean value according to the brightness distribution vector of each pixel point and the brightness distribution vector of each pixel point in the suspected non-bump pixel cluster;
obtaining the reliability by using the difference value of the first similarity mean value and the second similarity mean value obtained by each pixel point and the similarity mean value variance of all position pixel points;
calculating the bulge probability of each pixel point according to the first similarity mean value, the second similarity mean value and the reliability obtained by each pixel point and the category to which the pixel point belongs;
and judging whether each pixel point is a bulge pixel or not according to the bulge probability value of each pixel point and the bulge probability threshold value.
2. The method for detecting the defects of the ship electrical fittings as claimed in claim 1, wherein the method for obtaining the similarity mean value of the pixel points at each position comprises the following steps:
calculating the cosine similarity of the brightness distribution vectors of the pixel points at the positions and the pixel points at other positions according to the brightness distribution vectors of the pixel points at the positions and the pixel points at other positions;
and obtaining the similarity mean value of the pixel points at each position according to the cosine similarity of the brightness distribution vectors of the pixel points at each position and the pixel points at other positions.
3. The method for detecting the defects of the ship electrical fittings as claimed in claim 1, wherein the method for obtaining the credibility comprises the following steps:
obtaining a similarity difference value of each pixel point according to the first similarity mean value and the second similarity mean value obtained by each pixel point and the pixel point;
and obtaining the reliability by using the similarity mean variance of all the position pixel points and the similarity difference of all the pixel points.
4. The method for detecting the defects of the ship electrical fittings as claimed in claim 1, wherein the expression of the bulge probability of each pixel point is as follows:
the expression of the bump probability of a suspected bump pixel is:
Figure 770033DEST_PATH_IMAGE002
in the formula:
Figure DEST_PATH_IMAGE003
indicating the first of a cluster of suspected bump pixels
Figure 731036DEST_PATH_IMAGE004
The similarity mean value of each pixel point and the similar pixels,
Figure DEST_PATH_IMAGE005
indicating the first of a cluster of suspected bump pixels
Figure 484578DEST_PATH_IMAGE004
The similarity mean value of each pixel point and non-similar pixels,
Figure 353177DEST_PATH_IMAGE006
the degree of confidence is expressed in the form of,
Figure DEST_PATH_IMAGE007
representing the similarity mean variance of all the position pixel points,
Figure 918019DEST_PATH_IMAGE008
indicating the first of a cluster of suspected bump pixels
Figure 174557DEST_PATH_IMAGE004
The bump probability of each pixel point;
the expression for the bump probability for a suspected non-bump pixel is:
Figure 637900DEST_PATH_IMAGE010
in the formula:
Figure DEST_PATH_IMAGE011
indicating a suspected non-bump pixel cluster
Figure 375436DEST_PATH_IMAGE012
The similarity mean value of each pixel point and the similar pixels,
Figure DEST_PATH_IMAGE013
indicating a suspected non-bump pixel cluster
Figure 545517DEST_PATH_IMAGE012
The similarity mean value of each pixel point and non-similar pixels,
Figure 520427DEST_PATH_IMAGE014
indicating a suspected non-bump pixel cluster
Figure 720333DEST_PATH_IMAGE012
The bump probability of each pixel.
5. The method for detecting the defects of the ship electrical accessories, according to claim 1, is characterized in that the method for judging whether each pixel point is a bulge pixel according to the bulge probability value of each pixel point and the bulge probability threshold value comprises the following steps:
if the bulge probability value of the pixel point is larger than the bulge probability threshold value, judging the pixel point as a bulge pixel;
and if the swelling probability value of the pixel point is less than or equal to the swelling probability threshold value, judging that the pixel point is a non-swelling pixel.
6. The method for detecting the defects of the ship electrical accessories according to claim 1, wherein the suspected bump pixel clustering is a clustering with a small number of pixels obtained after clustering the pixels at all positions; the suspected non-bulge pixel clustering is a clustering with a large number of pixel points obtained after clustering the pixel points at all positions.
CN202210791378.2A 2022-07-07 2022-07-07 Ship electrical accessory defect detection method Active CN114862854B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210791378.2A CN114862854B (en) 2022-07-07 2022-07-07 Ship electrical accessory defect detection method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210791378.2A CN114862854B (en) 2022-07-07 2022-07-07 Ship electrical accessory defect detection method

Publications (2)

Publication Number Publication Date
CN114862854A CN114862854A (en) 2022-08-05
CN114862854B true CN114862854B (en) 2022-09-02

Family

ID=82625706

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210791378.2A Active CN114862854B (en) 2022-07-07 2022-07-07 Ship electrical accessory defect detection method

Country Status (1)

Country Link
CN (1) CN114862854B (en)

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP5616974B2 (en) * 2010-11-16 2014-10-29 本田技研工業株式会社 In-vehicle camera displacement detection device
CN102509075B (en) * 2011-10-19 2013-07-24 北京国铁华晨通信信息技术有限公司 Remnant object detection method and device
CN109668832A (en) * 2019-01-30 2019-04-23 江苏理工学院 Cellular li-ion battery surface bulge detection device and its method based on subregion light source
CN111964924A (en) * 2020-08-12 2020-11-20 合肥富煌君达高科信息技术有限公司 Tire bulge detection system based on high-speed camera and multi-line laser

Also Published As

Publication number Publication date
CN114862854A (en) 2022-08-05

Similar Documents

Publication Publication Date Title
CN107679535B (en) Automatic reading identification system and method for pointer type water meter based on template matching
CN106709500B (en) Image feature matching method
CN116091504B (en) Connecting pipe connector quality detection method based on image processing
CN111476246B (en) Robust and efficient intelligent reading method for pointer instrument applied to complex environment
CN116721107B (en) Intelligent monitoring system for cable production quality
CN111079596A (en) System and method for identifying typical marine artificial target of high-resolution remote sensing image
CN112734729B (en) Water gauge water level line image detection method and device suitable for night light supplement condition and storage medium
CN114882043B (en) Injection molding part defect positioning method and system based on image processing
CN115082508B (en) Ocean buoy production quality detection method
CN112380973A (en) Traffic signal lamp identification method and system
CN114067147B (en) Ship target confirmation method based on local shape matching
CN108694415B (en) Image feature extraction method and device and water source image classification method and device
CN110516565A (en) A kind of sea ship recognition methods based on Zernike and SVM
CN114862854B (en) Ship electrical accessory defect detection method
CN113781511A (en) Conveyor belt edge wear detection method and device, computer equipment and storage medium
CN113781523A (en) Football detection tracking method and device, electronic equipment and storage medium
CN116612306B (en) Computer vision-based intelligent flange plate alignment method and system
CN111104857A (en) Identity recognition method and system based on gait energy diagram
CN115409803A (en) Oil level identification method, system and medium for transformer in power distribution room
Jia et al. A novel moving cast shadow detection of vehicles in traffic scene
CN110853059B (en) Image identification method for target ring number
Abdusalomov et al. Robust shadow removal technique for improving image enhancement based on segmentation method
CN112990064A (en) Dial pointer identification method based on color segmentation and probability model
CN109886133A (en) A kind of ship detection method and system based on remote sensing image
Kang et al. Research on automatic recognition algorithm of pointer meter based on machine vision

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