CN117333825B - Cable bridge monitoring method based on computer vision - Google Patents
Cable bridge monitoring method based on computer vision Download PDFInfo
- Publication number
- CN117333825B CN117333825B CN202311628198.3A CN202311628198A CN117333825B CN 117333825 B CN117333825 B CN 117333825B CN 202311628198 A CN202311628198 A CN 202311628198A CN 117333825 B CN117333825 B CN 117333825B
- Authority
- CN
- China
- Prior art keywords
- image
- channel
- channel image
- pixel point
- hyperspectral
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 52
- 238000012544 monitoring process Methods 0.000 title claims abstract description 17
- HCHKCACWOHOZIP-UHFFFAOYSA-N Zinc Chemical compound [Zn] HCHKCACWOHOZIP-UHFFFAOYSA-N 0.000 claims abstract description 53
- 229910052725 zinc Inorganic materials 0.000 claims abstract description 53
- 239000011701 zinc Substances 0.000 claims abstract description 53
- 238000007747 plating Methods 0.000 claims abstract description 50
- 230000005856 abnormality Effects 0.000 claims abstract description 41
- 238000004458 analytical method Methods 0.000 claims abstract description 8
- 238000012545 processing Methods 0.000 claims abstract description 7
- 230000002159 abnormal effect Effects 0.000 claims description 55
- 230000009467 reduction Effects 0.000 claims description 19
- 239000013598 vector Substances 0.000 claims description 13
- 230000008569 process Effects 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000005096 rolling process Methods 0.000 claims description 3
- 238000010586 diagram Methods 0.000 claims description 2
- 229910052751 metal Inorganic materials 0.000 description 6
- 239000002184 metal Substances 0.000 description 6
- 238000001228 spectrum Methods 0.000 description 5
- 150000002739 metals Chemical class 0.000 description 3
- 239000011248 coating agent Substances 0.000 description 2
- 238000000576 coating method Methods 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 230000003595 spectral effect Effects 0.000 description 2
- 241001270131 Agaricus moelleri Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000005246 galvanizing Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- JEIPFZHSYJVQDO-UHFFFAOYSA-N iron(III) oxide Inorganic materials O=[Fe]O[Fe]=O JEIPFZHSYJVQDO-UHFFFAOYSA-N 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 230000001681 protective effect Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 230000003746 surface roughness Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- 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
- 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
- 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
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- 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
Abstract
The invention relates to the field of image processing, in particular to a cable bridge monitoring method based on computer vision, which comprises the steps of obtaining the gray level abnormality degree of a hyperspectral image channel image on the surface of each cable galvanized bridge, determining hyperspectral images with zinc plating unevenness according to the gray level abnormality degree, obtaining the abnormality degree of each pixel point in each channel image, calculating the probability that the pixel point is a zinc plating abnormality pixel point, obtaining zinc plating abnormality pixel points according to the probability, carrying out mean shift clustering on the zinc plating abnormality pixel points in each channel image, obtaining the optimal reference weight of each channel image according to each clustering result and the gray level abnormality degree, and carrying out gray level and connected domain analysis on the hyperspectral images according to the optimal reference weight to obtain the zinc plating uneven area on the surface of the cable galvanized bridge. According to the invention, the zinc plating uneven area on the surface of the cable zinc plating bridge is accurately and reliably monitored by extracting the characteristics in the image.
Description
Technical Field
The application relates to the field of computer vision, in particular to a cable bridge monitoring method based on computer vision.
Background
The cable bridge is a product derived for protecting the cable, is a protective shell of the cable and is used for preventing the cable from being damaged by external factors, and mainly comprises a bracket, a bracket arm and a mounting accessory. All parts of the cable bridge are required to be galvanized, so that the galvanized cable bridge is more attractive, can play a role in rust prevention, has a good protection effect on materials which are always exposed outdoors, and can prolong the service life and improve the durability of the cable bridge. However, many factors such as the metal composition of the bridge, the surface roughness, the geometry of the workpiece, the hot dip galvanizing process, etc. affect the thickness of the galvanized layer, which if uneven, would reduce the corrosion resistance of the bridge and easily cause defects on the bridge surface. The more uniform the thickness of the zinc coating on the bridge frame, the better the bridge frame quality, so that the uniformity of the zinc coating of the cable bridge frame needs to be monitored to ensure the production quality of the cable bridge frame.
Because the color change of the metal on the surface of the bridge is small in the images before and after the bridge galvanization, which are acquired by the traditional camera, the common camera is difficult to distinguish the fine differences of different positions, and the thickness of the galvanized layer at different positions is difficult to evaluate.
Therefore, according to the technical scheme, different characteristic spectrums exist according to different metals, the spectrum intensity and the content of the metals are also in a definite relation, the hyperspectral camera is used for collecting the cable bridge surface image, and the fine difference of each position in the bridge galvanized image is obtained by analyzing the spectrum information in the image, so that the area with uneven galvanization is obtained, and the monitoring of the galvanization quality of the bridge surface is realized.
Disclosure of Invention
The invention provides a cable bridge monitoring method based on computer vision, which solves the problem of inaccurate monitoring of the galvanization quality of the surface of a cable bridge, and adopts the following technical scheme:
the invention provides a cable bridge monitoring method based on computer vision, which comprises the following steps:
collecting hyperspectral images of the surface of the cable zinc-plated bridge frame, and performing dimension reduction treatment to obtain dimension-reduced hyperspectral images;
obtaining the gray level abnormality degree of each channel image according to the difference of the gray level value of each pixel point in each channel image of the hyperspectral image after dimension reduction relative to the gray level value in the channel image;
determining hyperspectral images with zinc plating unevenness according to the gray level abnormality degree of each channel image;
obtaining the abnormal degree of each pixel point in each channel image according to the gray value of each pixel point in each channel image in the hyperspectral image with zinc plating unevenness;
calculating the probability that each pixel point is a galvanization abnormal pixel point according to the abnormality degree of the pixel point in each channel image in the hyperspectral image with galvanization non-uniformity;
determining galvanization abnormal pixel points in the hyperspectral image with galvanization non-uniformity according to the probability and the probability threshold;
performing mean shift clustering on the galvanization abnormal pixel points in each channel image of the hyperspectral image with the galvanization non-uniformity;
obtaining optimal reference weight of each channel image according to the number of pixel points in each clustering result, the abnormal degree of galvanized abnormal pixel points in the channel image and the gray level abnormal degree of the channel image;
and carrying out graying treatment and connected domain analysis on the hyperspectral image according to the optimal reference weight of each channel image to obtain a zinc plating uneven area on the surface of the cable zinc plating bridge.
Further, the method for dimension reduction treatment comprises the following steps:
acquiring a gray level histogram of each channel image normalized by the hyperspectral image;
taking the ratio of the maximum gray value and the minimum gray value in the gray histogram corresponding to each channel image and each gray value in the gray range as one dimension of a vector, and converting the gray histogram of each channel image into a vector with the same dimension as the number of the gray values in the gray histogram;
dividing channel images with the same maximum gray value and the same minimum gray value in each channel image into a group, and calculating cosine similarity between vectors corresponding to the channel images in the same group;
the channel images corresponding to the two vectors with the cosine similarity value larger than or equal to the threshold value are mutually redundant channel images, the channel image with the smallest dimension in the images which are mutually redundant channels is reserved, and other channel images are removed;
and sequentially processing all images which are mutually redundant channels, so as to realize dimension reduction of the hyperspectral image.
Further, the method for determining the hyperspectral image with zinc plating unevenness comprises the following steps:
acquiring the gray level abnormality degree of each channel image, and obtaining a gray level abnormality degree average value as a hyperspectral image, wherein the possibility of zinc plating unevenness exists;
the likelihood is compared with a likelihood threshold, and when the likelihood is greater than the threshold, the hyperspectral image is a hyperspectral image in which there is zinc plating unevenness.
Further, the method for acquiring the galvanization abnormal pixel points in the hyperspectral image comprises the following steps:
obtaining the abnormal degree of each pixel point in the hyperspectral image in each channel image:
in the method, in the process of the invention,is->Gray value with maximum duty ratio of each channel image,/->Is->The pixel point is at the +.>Gray values in the individual channel images, +.>Is->The pixel point is at the +.>Degree of abnormality in the individual channel images;
calculating the probability that each pixel point in the hyperspectral image is a galvanization abnormal pixel point, wherein the calculation formula is as follows:
in the method, in the process of the invention,the number of channel images of the hyperspectral image after dimension reduction is +.>Is->The probability that each pixel point is a galvanization abnormal pixel point;
when (when)When the probability threshold is smaller than +.>The pixel points are noise points, otherwise, the first pixel point is the first pixel point>The pixel points are galvanization abnormal pixel points in the hyperspectral image.
Further, the method for obtaining the optimal reference weight of each channel image is as follows:
taking the gray level abnormality degree of each channel image as an initial reference weight of the channel image;
mean shift clustering coordinates of galvanized outliers in each channel image, where the firstThe number of mean shift clustering results obtained from each channel image is +.>Each cluster result corresponds to an uneven area;
the optimal reference weight calculation formula for each channel image is:
in the method, in the process of the invention,is->Optimal reference weights for the individual channel images, +.>Is->Initial reference weights of the individual channel images, +.>Is->The>The number of pixel points contained in the clustering result, namely the area of the uneven area in the clustering result,/->For the total number of clustering results, +.>Is->In the individual channel image +.>The +.>The galvanization abnormal pixel point is at the +.>Degree of abnormality in each channel image.
Further, the method for acquiring the zinc plating uneven area on the surface of the cable zinc plating bridge comprises the following steps:
accumulating and summing gray values of pixel points in each channel image by utilizing the optimal reference weight of each channel image, and graying the hyperspectral image to obtain a gray image of the hyperspectral image;
and (3) carrying out connected domain analysis on the gray level diagram of the hyperspectral image by using a Seed-rolling algorithm, wherein the obtained connected domain is the galvanized non-uniform area on the surface of the cable galvanized bridge.
Further, the method for obtaining the abnormality degree includes:
and obtaining the abnormal degree of each pixel point in each channel image according to the difference value between the gray value of each pixel point in the hyperspectral image with zinc plating unevenness in each channel image and the gray value with the largest duty ratio in the channel image.
Further, the method for acquiring the gray level abnormality degree comprises the following steps:
and obtaining the gray level abnormality degree of each channel image according to the variance of the gray level value of each pixel point in each channel image of the hyperspectral image after dimension reduction relative to the average gray level value in the channel image.
Further, the likelihood threshold is set to 0.1.
Further, the probability threshold is set to 0.6.
The beneficial effects of the invention are as follows: the method comprises the steps of obtaining gray level anomaly degree of each channel image by carrying out dimension reduction treatment on a hyperspectral image on the surface of a cable zinc plating bridge based on hyperspectral images, determining hyperspectral images with zinc plating unevenness according to the gray level anomaly degree, obtaining the anomaly degree of each pixel point in each channel image according to the difference value of the gray level value of each pixel point in each channel image and the gray level value with the largest proportion in the channel image, calculating the probability that the pixel point is the zinc plating anomaly pixel point according to the anomaly degree, determining the zinc plating anomaly pixel point in the hyperspectral image with zinc plating unevenness according to the probability and a threshold value, carrying out mean shift clustering on the zinc plating anomaly pixel point in each channel image with zinc plating unevenness according to the number of the pixel point in each clustering result and the anomaly degree of the zinc plating anomaly pixel point in the channel image and the anomaly degree of the channel image, obtaining optimal reference weight of each channel image, carrying out accurate treatment on a hyperspectral image and a zinc plating bridge according to the optimal reference weight of each channel image, carrying out accurate analysis on a zinc plating area, and obtaining a zinc plating bridge surface area.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only some embodiments of the invention, and that other drawings can be obtained according to these drawings without inventive faculty for a person skilled in the art.
Fig. 1 is a schematic flow chart of a cable bridge monitoring method based on computer vision.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
An embodiment of a cable bridge monitoring method based on computer vision of the present invention, as shown in fig. 1, includes:
step one: and collecting hyperspectral images of the surface of the cable zinc-plated bridge.
Step two: and (5) performing dimension reduction treatment to obtain a dimension-reduced hyperspectral image.
And collecting the hyperspectral images of the surface of the galvanized bridge by using a hyperspectral camera, and combining according to the similarity of the images of all the channels of the hyperspectral images, so that the redundancy of the hyperspectral images is reduced.
The main scene of this embodiment is: under uniform illumination, a hyperspectral camera is arranged right above a galvanized bridge, the focal length of the camera is adjusted, the visual field range of the camera is the width of the bridge, hyperspectral images corresponding to the surface of the bridge are collected, the hyperspectral images are processed, and the non-uniform galvanized area is determined according to characteristic information in the images, so that monitoring of the cable bridge is realized.
The specific method for the dimension reduction treatment comprises the following steps:
(1) Acquiring normalized gray histograms in each channel image, wherein the horizontal axis in the gray histograms represents gray values and the vertical axis represents the proportion of each gray value;
(2) Obtaining the maximum gray value and the minimum gray value in the gray histogram corresponding to each channel image, and taking the proportion of each gray value in the gray range as one dimension of the vector, thereby converting each gray histogram into a vector with the same dimension as the number of the gray values in the gray histogram, and then the firstThe gray scale characteristics of the individual channel images can be expressed as +.>Wherein->Is->Minimum gray value and maximum gray value in the individual channel images, +.>Is->Vectors corresponding to the channel images;
(3) Dividing channel images with the same maximum gray value and the same minimum gray value into a group according to each channel image, calculating cosine similarity between vectors corresponding to the channel images in the same group, and mutually redundant channel the channel images corresponding to two vectors with the cosine similarity value larger than or equal to a similarity threshold value, wherein the similarity threshold value is 0.95 in the embodiment;
(4) Reserving channel images with minimum dimension in images of mutually redundant channels, removing other channel images, sequentially processing all the images of mutually redundant channels, realizing dimension reduction of hyperspectral images, obtaining a channel dimension sequence reserved after dimension reduction, and recording the number of the channel images after dimension reduction as。
It should be noted that, because the hyperspectral image is subdivided based on spectral bands, the spectral resolution is high, the hyperspectral image contains more bands, each pixel point in the image corresponds to a plurality of channels, so that each pixel point contains pixel values in a plurality of dimensions, but because the bands of the hyperspectral image have strong correlation, the correlation coefficient between the spectrums of the image is large, hyperspectral redundant information stacking is easy to be caused, and the redundancy increases along with the increase of the number of imaging bands and the imaging resolution, and the hyperspectral image has typical high redundancy characteristics. In order to reduce the calculation amount, channels with larger similarity between the images of each channel in the hyperspectral image are required to be combined, so that the dimension of the hyperspectral image is reduced.
Step three: gray level abnormality degree of each channel image.
Step four: there is a hyperspectral image of zinc plating unevenness.
The hyperspectral image of the zinc plating non-uniformity phenomenon is primarily judged, because the zinc plating non-uniformity phenomenon is not present in all zinc plating bridge images, the zinc plating uniformity degree needs to be primarily judged to avoid unnecessary operation, and the gray values in each channel image are the same when zinc plating is uniform, so that the primary judgment of the zinc plating uniformity degree can be firstly carried out according to the difference degree of the gray values in each channel.
The method for determining the hyperspectral image with zinc plating unevenness comprises the following steps:
(1) Acquiring the gray level abnormality degree of each channel image: since the gray values in each channel are uniform when the galvanization is uniform, the variance of all gray values and the proportion thereof in each gray histogram relative to the average gray value of the corresponding channel image is calculated, and the abnormal degree of the gray in the channel image is represented by the normalized result of the obtained variance, wherein the firstThe gray level abnormality degree of each channel image is recorded as +.>；
(2) Calculating a gray level abnormity degree average value: because the gray scales of different metals in partial channels are similar, whether the surface of the bridge frame has a galvanized uneven area cannot be judged according to the gray scale abnormality degree in a single channel image, and comprehensive evaluation is required to be performed by integrating the gray scale abnormality degrees of a plurality of channel images, so that the average value of the gray scale abnormality degrees of all the channel images is calculated, and the average value is used as a hyperspectral image to have the possibility of galvanization uneven；
(3) According toAnd (3) judging: possibility +.>And likelihood threshold->In contrast, when->Is greater than->In this case, the hyperspectral image is a hyperspectral image in which zinc plating unevenness exists, and the probability threshold value in this embodiment is 0.1.
Step five: the degree of abnormality of each pixel point in the channel image.
Step six: each pixel point is the probability of a galvanization abnormal pixel point.
And analyzing the abnormal degree of each pixel point in each channel, and comprehensively judging to obtain galvanization abnormal pixel points.
The method for acquiring the galvanization abnormal pixel points comprises the following steps:
(1) Obtaining the abnormal degree of each pixel point in the hyperspectral image in each channel image: because most areas in the galvanized image are areas with uniform galvanization, according to the gray value with the largest proportion in each channel image as the reference gray value of the corresponding channel image, taking the difference value of the gray value of each pixel point relative to the reference gray value of each channel as the abnormal degree of the pixel point in each channel;
the calculation formula of the degree of abnormality of each pixel point in each channel image is as follows:
in the method, in the process of the invention,is->Gray value with maximum duty ratio of each channel image,/->Is->The pixel point is at the +.>Gray values in the individual channel images, +.>Is->The pixel point is at the +.>Degree of abnormality in the individual channel images;
(2) Calculating the probability that each pixel point in the hyperspectral image is a galvanization abnormal pixel point: in order to eliminate the interference of noise points, the possibility that the pixel point is a galvanized abnormal point needs to be judged by combining the abnormal degree in other channel images, and the first spectrum image after dimension reductionThe probability that each pixel point is a galvanized outlier is +.>The calculation formula is as follows:
in the method, in the process of the invention,the number of channel images of the hyperspectral image after dimension reduction is +.>Is->Personal imageThe pixel points are the probability of abnormal galvanization pixel points;
(3) Judging according to the probability and the probability threshold value: when (when)When the probability threshold is smaller than +.>The pixel points are noise points, otherwise, the first pixel point is the first pixel point>The pixels are abnormal pixels in the hyperspectral image, and the probability threshold in the embodiment is 0.6.
The more channels of the image, the more likely it is to generate noise. In the same channel image, gray level difference exists between noise and metal zinc, but gray level difference exists in abnormal positions generated by uneven galvanization, so that suspected abnormal points obtained according to the gray level difference of pixel points in a single channel contain abnormal points and noise points, namely, the abnormal points and the noise points cannot be distinguished according to the difference condition between gray values in the single channel image.
Step seven: optimal reference weights for each channel image.
According to the distribution condition of the galvanized abnormal pixel points, the reference weight of each channel image is adjusted to obtain the optimal reference weight of the image graying, because the weight of each channel is an empirical value distributed based on the sensitivity degree of human eyes to three colors of red, green and blue when the RGB camera converts the image into the gray image, but when more channels exist in the image, no empirical value can be used for the graying processing of the hyperspectral image, and the abnormal degree of each channel is different, namely, the distributed weights for different channels are different when the hyperspectral image is subjected to the graying processing, so that the subsequent extraction process of the uneven region tends to be more accurate, the channel image which can more highlight the abnormal region tends to be given a larger weight, and therefore, the invention adjusts the reference weight of the channel according to the abnormal degree of the pixel points in different channels, namely, the optimal reference weight.
The method for acquiring the optimal reference weight of each channel image comprises the following steps:
(1) Taking the gray level abnormality degree of each channel image obtained in the second step as the initial reference weight of the channel image, then the first stepThe initial reference weight of the individual channel image is +.>：
Wherein,is->Gray level abnormality degree of individual channel image, when +.>When the channel images are formed by fusing the images which are mutually redundant channels, the gray level abnormality degree is the average value of the gray level abnormality degrees of the images of the redundant channels.
(2) Mean shift clustering of coordinates of galvanized outliers in each channel image, wherein the firstThe number of mean shift clustering results obtained from each channel image is +.>Each cluster result corresponds to an uneven area;
then the firstThe optimal reference weights for the individual channel images can be expressed as:
wherein,is->The>The number of pixel points contained in the clustering result can also represent the uneven area in the clustering result, < >>For the total number of clustering results, +.>Is->In the individual channel image +.>The>Degree of abnormality of individual galvanization abnormal points within the channel image.
It should be noted that, when each galvanized uneven area in the channel image may show a more complete area, that is, a larger area, the reference degree to the channel image is larger; the invention aims to make the galvanized uneven area simpler in the subsequent segmentation process, so that the uneven area needs to be more obvious, the obvious degree of the uneven area can be represented by the abnormal degree of each galvanized abnormal point in the area, and the greater the abnormal degree of the uneven area is, the greater the reference degree of the channel image is, therefore, the step needs to comprehensively evaluate the area of the abnormal area and the abnormal degree of the area which can be displayed in the channel image when the reference degree of the channel image is adjusted.
Step eight: and (5) carrying out hyperspectral image graying treatment and connected domain analysis.
Step nine: zinc plating uneven area of the surface of the cable zinc plating bridge.
And acquiring a gray level image corresponding to the hyperspectral image according to the reference weight of each channel, and extracting an abnormal region on the surface of the cable bridge.
The method for acquiring the zinc plating uneven area comprises the following steps:
(1) The gray values of pixel points in each channel image are accumulated and summed by combining the optimal reference weights corresponding to each channel, so that the hyperspectral image is subjected to gray processing;
(2) And (3) carrying out connected domain analysis on the image by using a Seed-rolling algorithm, wherein the obtained connected region is the non-uniform galvanized region on the surface of the cable bridge.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.
Claims (6)
1. A method for monitoring a cable bridge based on computer vision, comprising:
collecting hyperspectral images of the surface of the cable zinc-plated bridge frame, and performing dimension reduction treatment to obtain dimension-reduced hyperspectral images;
obtaining the gray level abnormality degree of each channel image according to the difference of the gray level value of each pixel point in each channel image of the hyperspectral image after dimension reduction relative to the gray level value in the channel image;
determining hyperspectral images with zinc plating unevenness according to the gray level abnormality degree of each channel image;
obtaining the abnormal degree of each pixel point in each channel image according to the gray value of each pixel point in each channel image in the hyperspectral image with zinc plating unevenness;
calculating the probability that each pixel point is a galvanization abnormal pixel point according to the abnormality degree of the pixel point in each channel image in the hyperspectral image with galvanization non-uniformity;
determining galvanization abnormal pixel points in the hyperspectral image with galvanization non-uniformity according to the probability and the probability threshold;
performing mean shift clustering on the galvanization abnormal pixel points in each channel image of the hyperspectral image with the galvanization non-uniformity;
obtaining optimal reference weight of each channel image according to the number of pixel points in each clustering result, the abnormal degree of galvanized abnormal pixel points in the channel image and the gray level abnormal degree of the channel image;
carrying out graying treatment and connected domain analysis on the hyperspectral image according to the optimal reference weight of each channel image to obtain a zinc plating uneven area on the surface of the cable zinc plating bridge;
the method for determining the hyperspectral image with zinc plating unevenness comprises the following steps:
acquiring the gray level abnormality degree of each channel image, and obtaining a gray level abnormality degree average value as a hyperspectral image, wherein the possibility of zinc plating unevenness exists;
comparing the probability with a probability threshold, wherein when the probability is greater than the threshold, the hyperspectral image is a hyperspectral image with zinc plating unevenness;
the method for acquiring the galvanization abnormal pixel points in the hyperspectral image comprises the following steps:
obtaining the abnormal degree of each pixel point in the hyperspectral image in each channel image:
in the method, in the process of the invention,is->Gray value with maximum duty ratio of each channel image,/->Is->The pixel point is at the +.>Gray values in the individual channel images, +.>Is->The pixel point is at the +.>Degree of abnormality in the individual channel images;
calculating the probability that each pixel point in the hyperspectral image is a galvanization abnormal pixel point, wherein the calculation formula is as follows:
in the method, in the process of the invention,the number of channel images of the hyperspectral image after dimension reduction is +.>Is->The probability that each pixel point is a galvanization abnormal pixel point;
when (when)When the probability threshold is smaller than +.>The pixel points are noise points, otherwise, the first pixel point is the first pixel point>The pixel points are galvanization abnormal pixel points in the hyperspectral image; the probability threshold is set to 0.6;
the method for acquiring the optimal reference weight of each channel image comprises the following steps:
taking the gray level abnormality degree of each channel image as an initial reference weight of the channel image;
mean shift clustering coordinates of galvanized outliers in each channel image, where the firstThe number of mean shift clustering results obtained from each channel image is +.>Each cluster result corresponds to an uneven area;
the optimal reference weight calculation formula for each channel image is:
in the method, in the process of the invention,is->Optimal reference weights for the individual channel images, +.>Is->Initial reference weights of the individual channel images, +.>Is->The>The number of pixel points contained in the clustering result, namely the area of the uneven area in the clustering result,/->For the total number of clustering results, +.>Is->In the individual channel image +.>The +.>The galvanization abnormal pixel point is at the +.>Degree of abnormality in each channel image.
2. The method for monitoring the cable bridge based on computer vision according to claim 1, wherein the method for dimension reduction treatment is as follows:
acquiring a gray level histogram of each channel image normalized by the hyperspectral image;
taking the ratio of the maximum gray value and the minimum gray value in the gray histogram corresponding to each channel image and each gray value in the gray range as one dimension of a vector, and converting the gray histogram of each channel image into a vector with the same dimension as the number of the gray values in the gray histogram;
dividing channel images with the same maximum gray value and the same minimum gray value in each channel image into a group, and calculating cosine similarity between vectors corresponding to the channel images in the same group;
the channel images corresponding to the two vectors with the cosine similarity value larger than or equal to the threshold value are mutually redundant channel images, the channel image with the smallest dimension in the images which are mutually redundant channels is reserved, and other channel images are removed;
and sequentially processing all images which are mutually redundant channels, so as to realize dimension reduction of the hyperspectral image.
3. The method for monitoring the cable bridge based on computer vision according to claim 1, wherein the method for obtaining the zinc plating uneven area on the surface of the cable zinc plating bridge is as follows:
accumulating and summing gray values of pixel points in each channel image by utilizing the optimal reference weight of each channel image, and graying the hyperspectral image to obtain a gray image of the hyperspectral image;
and (3) carrying out connected domain analysis on the gray level diagram of the hyperspectral image by using a Seed-rolling algorithm, wherein the obtained connected domain is the galvanized non-uniform area on the surface of the cable galvanized bridge.
4. The method for monitoring a cable bridge according to claim 1, wherein said method for obtaining the degree of anomaly comprises:
and obtaining the abnormal degree of each pixel point in each channel image according to the difference value between the gray value of each pixel point in the hyperspectral image with zinc plating unevenness in each channel image and the gray value with the largest duty ratio in the channel image.
5. The method for monitoring a cable bridge based on computer vision according to claim 1, wherein the method for obtaining the gray level anomaly degree comprises the following steps:
and obtaining the gray level abnormality degree of each channel image according to the variance of the gray level value of each pixel point in each channel image of the hyperspectral image after dimension reduction relative to the average gray level value in the channel image.
6. The computer vision based cable bridge monitoring method according to claim 1, wherein said probability threshold is set to 0.1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311628198.3A CN117333825B (en) | 2023-12-01 | 2023-12-01 | Cable bridge monitoring method based on computer vision |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311628198.3A CN117333825B (en) | 2023-12-01 | 2023-12-01 | Cable bridge monitoring method based on computer vision |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117333825A CN117333825A (en) | 2024-01-02 |
| CN117333825B true CN117333825B (en) | 2024-02-23 |
Family
ID=89279647
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311628198.3A Active CN117333825B (en) | 2023-12-01 | 2023-12-01 | Cable bridge monitoring method based on computer vision |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117333825B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117593651B (en) * | 2024-01-18 | 2024-04-05 | 四川交通职业技术学院 | Tunnel crack segmentation recognition method |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115082488A (en) * | 2022-08-23 | 2022-09-20 | 南通浩盛汽车科技有限公司 | Surface feather mark control method and device in automobile part galvanizing process |
| CN115184380A (en) * | 2022-08-01 | 2022-10-14 | 国网上海市电力公司 | Printed circuit board welding spot abnormity detection method based on machine vision |
| CN115294102A (en) * | 2022-09-26 | 2022-11-04 | 如东延峰钢结构有限公司 | Stainless steel product abnormity identification method based on machine vision |
| CN115345885A (en) * | 2022-10-19 | 2022-11-15 | 南通鹏宝运动用品有限公司 | Method for detecting appearance quality of metal fitness equipment |
| CN116448769A (en) * | 2023-05-10 | 2023-07-18 | 南京林业大学 | Multi-mode information fusion plate defect detection system and detection method thereof |
| EP4250224A1 (en) * | 2022-03-25 | 2023-09-27 | Primetals Technologies Japan, Ltd. | Method of correcting image, method of detecting abnormality, image correcting apparatus, and abnormality detecting apparatus |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB201900437D0 (en) * | 2019-01-11 | 2019-02-27 | Axial Medical Printing Ltd | Axial3d big book 2 |
| CN114267291B (en) * | 2020-09-16 | 2023-05-12 | 京东方科技集团股份有限公司 | Gray scale data determination method, device, equipment and screen driving plate |
-
2023
- 2023-12-01 CN CN202311628198.3A patent/CN117333825B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP4250224A1 (en) * | 2022-03-25 | 2023-09-27 | Primetals Technologies Japan, Ltd. | Method of correcting image, method of detecting abnormality, image correcting apparatus, and abnormality detecting apparatus |
| CN115184380A (en) * | 2022-08-01 | 2022-10-14 | 国网上海市电力公司 | Printed circuit board welding spot abnormity detection method based on machine vision |
| CN115082488A (en) * | 2022-08-23 | 2022-09-20 | 南通浩盛汽车科技有限公司 | Surface feather mark control method and device in automobile part galvanizing process |
| CN115294102A (en) * | 2022-09-26 | 2022-11-04 | 如东延峰钢结构有限公司 | Stainless steel product abnormity identification method based on machine vision |
| CN115345885A (en) * | 2022-10-19 | 2022-11-15 | 南通鹏宝运动用品有限公司 | Method for detecting appearance quality of metal fitness equipment |
| CN116448769A (en) * | 2023-05-10 | 2023-07-18 | 南京林业大学 | Multi-mode information fusion plate defect detection system and detection method thereof |
Non-Patent Citations (3)
| Title |
|---|
| Optimal combining fusion on degraded compressed sensing image reconstruction;Islam, SR;《SIGNAL PROCESSING-IMAGE COMMUNICATION》;20170412;第52卷;173-182 * |
| 基于机器视觉的锚杆异常快速检测方法;王昱栋;《工矿自动化》;20210430(第4期);13-18 * |
| 镀锌带钢设备裂缝图像的分割技术研究;毕军涛;温雪;丁喜纲;孙授卿;;世界有色金属;20160325(06);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117333825A (en) | 2024-01-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN117333825B (en) | Cable bridge monitoring method based on computer vision | |
| CN116091499B (en) | Abnormal paint production identification system | |
| CN108876768B (en) | Shadow defect detection method for light guide plate | |
| CN105891230B (en) | Fruit appearance detection method based on spectral image analysis | |
| US20100111362A1 (en) | Method for detecting shadow of object | |
| CN115841434A (en) | Infrared image enhancement method for gas concentration analysis | |
| CN113935666B (en) | Building decoration wall tile abnormity evaluation method based on image processing | |
| CN116452598A (en) | Axle production quality rapid detection method and system based on computer vision | |
| CN115330646B (en) | Metal component inspection method for fitness equipment paint spraying process | |
| CN115330796B (en) | Copper wire tinning defect identification method | |
| CN116740058A (en) | Quality detection method for solid state disk matched wafer | |
| CN114332081B (en) | Textile surface abnormity determination method based on image processing | |
| CN113744326B (en) | Fire detection method based on seed region growth rule in YCRCB color space | |
| Feng et al. | Research on defect detection method for high-reflective-metal surface based on high dynamic range imaging | |
| Petrellis | Plant disease diagnosis with color normalization | |
| CN112884795A (en) | Power transmission line inspection foreground and background segmentation method based on multi-feature significance fusion | |
| CN116721096B (en) | New energy harness quality online detection method based on artificial intelligence | |
| CN117011291B (en) | Watch shell quality visual detection method | |
| CN116630332B (en) | PVC plastic pipe orifice defect detection method based on image processing | |
| CN108230334B (en) | High-concentration wind-blown sand image segmentation method based on gray threshold | |
| CN116883412A (en) | Graphene far infrared electric heating equipment fault detection method | |
| CN116402810A (en) | Image processing-based lubricating oil anti-abrasive particle quality detection method | |
| CN115115622B (en) | Punching press terminal visual detection device | |
| JP6573798B2 (en) | Image processing apparatus and image processing method | |
| CN116758447A (en) | Digital twinning-based substation equipment fault analysis system |
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 |