CN116977853A - X-ray image-based transmission line crimping defect identification method and device - Google Patents
X-ray image-based transmission line crimping defect identification method and device Download PDFInfo
- Publication number
- CN116977853A CN116977853A CN202310948046.5A CN202310948046A CN116977853A CN 116977853 A CN116977853 A CN 116977853A CN 202310948046 A CN202310948046 A CN 202310948046A CN 116977853 A CN116977853 A CN 116977853A
- Authority
- CN
- China
- Prior art keywords
- ray image
- preset
- image
- graying
- optimized
- 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.)
- Pending
Links
- 230000007547 defect Effects 0.000 title claims abstract description 71
- 238000002788 crimping Methods 0.000 title claims abstract description 45
- 238000000034 method Methods 0.000 title claims abstract description 45
- 230000005540 biological transmission Effects 0.000 title claims abstract description 41
- 210000002569 neuron Anatomy 0.000 claims abstract description 36
- 238000012847 principal component analysis method Methods 0.000 claims abstract description 31
- 230000007246 mechanism Effects 0.000 claims abstract description 30
- 230000009466 transformation Effects 0.000 claims abstract description 29
- 239000002245 particle Substances 0.000 claims abstract description 25
- 238000012545 processing Methods 0.000 claims description 30
- 238000012549 training Methods 0.000 claims description 29
- 239000011159 matrix material Substances 0.000 claims description 19
- 230000006835 compression Effects 0.000 claims description 16
- 238000007906 compression Methods 0.000 claims description 16
- 238000007781 pre-processing Methods 0.000 claims description 15
- 238000003708 edge detection Methods 0.000 claims description 10
- 238000000513 principal component analysis Methods 0.000 claims description 10
- 238000005457 optimization Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000004140 cleaning Methods 0.000 claims description 7
- 238000013507 mapping Methods 0.000 claims description 7
- 230000011218 segmentation Effects 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 6
- 230000003044 adaptive effect Effects 0.000 claims description 5
- 230000010339 dilation Effects 0.000 claims description 3
- 238000010408 sweeping Methods 0.000 claims description 2
- 230000006870 function Effects 0.000 description 19
- 230000008569 process Effects 0.000 description 13
- 238000001228 spectrum Methods 0.000 description 6
- 238000001514 detection method Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 238000003860 storage Methods 0.000 description 4
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000003672 processing method Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000005855 radiation Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 241000519995 Stachys sylvatica Species 0.000 description 1
- 230000003321 amplification Effects 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000013441 quality evaluation Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000000547 structure data Methods 0.000 description 1
- 238000006467 substitution reaction Methods 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
- G06V20/10—Terrestrial scenes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/09—Supervised learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/30—Noise filtering
-
- 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
- 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
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
-
- 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
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Image Analysis (AREA)
Abstract
The application discloses a transmission line crimping defect identification method and device based on X-ray images, wherein the method comprises the following steps: respectively carrying out graying treatment and denoising treatment on an original X-ray image of a strain clamp of a power transmission line based on a preset self-adaptive graying algorithm and a preset FHN neuron parallel array model to obtain an optimized ray image, wherein the preset self-adaptive graying algorithm comprises a power law transformation mechanism, and the preset FHN neuron parallel array model is constructed based on a stochastic resonance mechanism; extracting features of the optimized radiographic image by adopting a principal component analysis method based on a kernel function to obtain a low-dimensional feature vector; and inputting the low-dimensional feature vector into an improved KNN model for classifying the crimping defects to obtain a crimping defect recognition result, wherein the improved KNN model is obtained by optimizing a particle swarm algorithm. The application can solve the technical problems that the prior art can not overcome the quality defect of the X-ray image and ensure the accuracy of the identification result of the image defect, thereby being incapable of meeting the actual application requirements.
Description
Technical Field
The application relates to the technical field of image recognition, in particular to a method and a device for recognizing crimping defects of a power transmission line based on an X-ray image.
Background
Crimping is an important material processing technology and is commonly used for connecting 110kV and above transmission lines with strain clamps and splicing sleeves. In the crimping process, the crimping defects are necessarily generated in the crimping process due to the influences of materials such as a power transmission line, a strain clamp, a splicing sleeve and the like, a crimping process, natural environment and the like. The quality of compression joint directly influences the construction quality and service life of a power transmission line, and even can cause catastrophic accidents, so that a great deal of economic losses and casualties are caused. Therefore, in order to ensure the stability and high efficiency of the power transmission line and reduce the probability of accidents, the quality detection of the compression joint of the strain clamp and the splicing sleeve is of profound significance. The ray spectrum is a visual display result of nondestructive quality detection and carries a large amount of visual information. However, due to the influence of many factors such as the process of the detection equipment and the detection method, various defects such as unobvious image brightness, contrast and gray level change, a large amount of noise interference and the like still exist in the ray spectrum, and the defects of the spectrum directly influence the accuracy of the subsequent crimping quality evaluation.
Because the existing X-ray image has unavoidable quality defects, the existing image defect identification technology based on a learning algorithm is difficult to extract the characteristic values of key parts, and the subsequent image processing process is provided with higher requirements, even the accuracy of a final detection result can be influenced, and the application requirements of the current stage can not be met.
Disclosure of Invention
The application provides a method and a device for identifying the crimping defect of a power transmission line based on an X-ray image, which are used for solving the technical problems that the quality defect of the X-ray image cannot be overcome, the accuracy of an image defect identification result is ensured, and the actual application requirement cannot be met.
In view of the above, a first aspect of the present application provides a method for identifying a crimping defect of a power transmission line based on an X-ray image, comprising:
respectively carrying out graying treatment and denoising treatment on an original X-ray image of a strain clamp of a power transmission line based on a preset self-adaptive graying algorithm and a preset FHN neuron parallel array model to obtain an optimized ray image, wherein the preset self-adaptive graying algorithm comprises a power law transformation mechanism, and the preset FHN neuron parallel array model is constructed based on a stochastic resonance mechanism;
extracting the characteristics of the optimized radiographic image by adopting an improved principal component analysis method to obtain a low-dimensional characteristic vector, wherein the improved principal component analysis method comprises a kernel function;
and inputting the low-dimensional feature vector into an improved KNN model for compression joint defect classification to obtain compression joint defect identification results, wherein the improved KNN model is obtained by optimizing a particle swarm algorithm.
Preferably, the performing gray-scale processing and denoising processing on the original X-ray image of the strain clamp of the power transmission line based on the preset adaptive gray-scale algorithm and the preset FHN neuron parallel array model to obtain an optimized ray image includes:
constructing a power exponent transformation formula corresponding to a preset self-adaptive graying algorithm based on a power law transformation mechanism;
performing self-adaptive graying treatment on an original X-ray image of the power transmission line through the power exponent transformation formula to obtain a graying ray image;
after a preset FHN neuron parallel array model is established based on a stochastic resonance mechanism, converting the gray-scale ray image into a binary sequence;
and denoising the binary sequence by adopting the preset FHN neuron parallel array model to obtain an optimized radiographic image.
Preferably, the gray processing and denoising processing are respectively performed on the original X-ray image of the strain clamp of the power transmission line based on the preset self-adaptive gray processing algorithm and the preset FHN neuron parallel array model to obtain an optimized ray image, and then the method further comprises the following steps:
the optimized radiographic image is subjected to preprocessing operations including edge detection, dilation, sweeping, segmentation and compression.
Preferably, the method for extracting features of the optimized radiographic image by using an improved principal component analysis method to obtain a low-dimensional feature vector, where the improved principal component analysis method includes a kernel function, including:
after the optimized radiographic image is expressed as an image data matrix, mapping the image data matrix to a high-dimensional feature space by improving a kernel function in a principal component analysis method to obtain a high-dimensional feature expression;
and carrying out principal component analysis and calculation based on the feature vector according to the high-dimensional feature expression to obtain a low-dimensional feature vector.
Preferably, the low-dimensional feature vector is input into an improved KNN model for classifying the crimping defects, so as to obtain a crimping defect recognition result, and the improved KNN model is optimized by a particle swarm algorithm and further comprises:
classifying and training the initial KNN model based on the number of the initialized adjacent samples and a preset training data set, and calculating classification errors;
taking the classification error as a fitness value, and updating the initial adjacent sample number to obtain an updated adjacent sample number;
and returning the step of classifying and training the initial KNN model based on the initialized adjacent sample number and the preset training data set based on the updated adjacent sample number until the iteration termination condition is reached, so as to obtain the optimal adjacent sample number and the improved KNN model.
A second aspect of the present application provides an X-ray image-based power line crimping defect recognition apparatus, including:
the image processing unit is used for respectively carrying out graying treatment and denoising treatment on an original X-ray image of a strain clamp of the power transmission line based on a preset self-adaptive graying algorithm and a preset FHN neuron parallel array model to obtain an optimized ray image, wherein the preset self-adaptive graying algorithm comprises a power law transformation mechanism, and the preset FHN neuron parallel array model is constructed based on a stochastic resonance mechanism;
the feature extraction unit is used for extracting the features of the optimized radiographic image by adopting an improved principal component analysis method to obtain a low-dimensional feature vector, wherein the improved principal component analysis method comprises a kernel function;
and the defect identification unit is used for inputting the low-dimensional feature vector into an improved KNN model for compression joint defect classification to obtain compression joint defect identification results, and the improved KNN model is obtained by optimizing a particle swarm algorithm.
Preferably, the image processing unit is specifically configured to:
constructing a power exponent transformation formula corresponding to a preset self-adaptive graying algorithm based on a power law transformation mechanism;
performing self-adaptive graying treatment on an original X-ray image of the power transmission line through the power exponent transformation formula to obtain a graying ray image;
after a preset FHN neuron parallel array model is established based on a stochastic resonance mechanism, converting the gray-scale ray image into a binary sequence;
and denoising the binary sequence by adopting the preset FHN neuron parallel array model to obtain an optimized radiographic image.
Preferably, the method further comprises:
and the preprocessing unit is used for preprocessing the optimized radiographic image, and the preprocessing operation comprises edge detection, expansion, cleaning, segmentation and compression.
Preferably, the feature extraction unit is specifically configured to:
after the optimized radiographic image is expressed as an image data matrix, mapping the image data matrix to a high-dimensional feature space by improving a kernel function in a principal component analysis method to obtain a high-dimensional feature expression;
and carrying out principal component analysis and calculation based on the feature vector according to the high-dimensional feature expression to obtain a low-dimensional feature vector.
Preferably, the method further comprises:
the classification training unit is used for carrying out classification training on the initial KNN model based on the number of the initialized adjacent samples and a preset training data set, and calculating classification errors;
a parameter updating unit, configured to take the classification error as a fitness value, and update the initial number of neighboring samples to obtain an updated number of neighboring samples;
and the iterative optimization unit is used for triggering the classification training unit based on the updated adjacent sample number until an iterative termination condition is reached, so as to obtain the optimal adjacent sample number and an improved KNN model.
From the above technical solutions, the embodiment of the present application has the following advantages:
the application provides a transmission line crimping defect identification method based on an X-ray image, which comprises the following steps: respectively carrying out graying treatment and denoising treatment on an original X-ray image of a strain clamp of a power transmission line based on a preset self-adaptive graying algorithm and a preset FHN neuron parallel array model to obtain an optimized ray image, wherein the preset self-adaptive graying algorithm comprises a power law transformation mechanism, and the preset FHN neuron parallel array model is constructed based on a stochastic resonance mechanism; extracting features of the optimized radiographic image by adopting an improved principal component analysis method to obtain a low-dimensional feature vector, wherein the improved principal component analysis method comprises a kernel function; and inputting the low-dimensional feature vector into an improved KNN model for classifying the crimping defects to obtain a crimping defect recognition result, wherein the improved KNN model is obtained by optimizing a particle swarm algorithm.
The transmission line crimping defect identification method based on the X-ray image provided by the application adopts an improved graying algorithm and a denoising processing method to carry out graying processing and denoising processing on an original X-ray image, and is used for improving the quality of the original X-ray image, reducing the pressure of subsequent image processing and improving the accuracy of defect identification from the source; in addition, in order to overcome the defect that the principal component analysis method lacks of nonlinear expression capability, a kernel function is introduced to improve the principal component analysis method, so that the nonlinear expression capability of the extracted low-dimensional feature vector is stronger; moreover, the improved KNN model optimized by the particle swarm optimization can further ensure the reliability of the crimping defect identification result. Therefore, the application can solve the technical problems that the prior art can not overcome the quality defect of the X-ray image and ensure the accuracy of the identification result of the image defect, thereby being incapable of meeting the actual application requirements.
Drawings
Fig. 1 is a schematic flow chart of a transmission line crimping defect identification method based on an X-ray image according to an embodiment of the present application;
fig. 2 is a schematic structural diagram of an X-ray image-based power transmission line crimping defect recognition device according to an embodiment of the present application;
FIG. 3 is a schematic diagram of an image effect after graying processing according to an embodiment of the present application;
fig. 4 is an exemplary diagram of a signal and a corresponding spectrum before denoising according to an embodiment of the present application;
fig. 5 is an exemplary diagram of a denoised signal and a corresponding spectrum according to an embodiment of the present application.
Detailed Description
In order to make the present application better understood by those skilled in the art, the following description will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings, and it is apparent that the described embodiments are only some embodiments of the present application, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
For easy understanding, referring to fig. 1, an embodiment of a method for identifying a crimping defect of a power transmission line based on an X-ray image according to the present application includes:
and step 101, respectively carrying out graying treatment and denoising treatment on an original X-ray image of a strain clamp of a power transmission line based on a preset self-adaptive graying algorithm and a preset FHN neuron parallel array model to obtain an optimized ray image, wherein the preset self-adaptive graying algorithm comprises a power law transformation mechanism, and the preset FHN neuron parallel array model is constructed based on a stochastic resonance mechanism.
Further, step 101 includes:
constructing a power exponent transformation formula corresponding to a preset self-adaptive graying algorithm based on a power law transformation mechanism;
performing self-adaptive graying treatment on an original X-ray image of the power transmission line through a power exponent transformation formula to obtain a graying ray image;
after constructing a preset FHN neuron parallel array model based on a stochastic resonance mechanism, converting the gray-scale ray image into a binary sequence;
and denoising the binary sequence by adopting a preset FHN neuron parallel array model to obtain an optimized radiographic image.
Because the original X-ray image of the power transmission line has the defects of unobvious image brightness, contrast, gray level change, large amount of noise interference and the like, the subsequent image processing process and defect identification accuracy can be directly influenced, and different image processing methods are designed to improve the image quality.
It should be noted that, constructing a power exponent transformation formula corresponding to a preset adaptive graying algorithm based on a power law transformation mechanism is expressed as follows:
G(x,y)＝ε×f(x,y) γ
wherein f (X, y) is an original X-ray image, epsilon is a conventional coefficient, and correction can be carried out according to the image effect, and the value is generally 1; gamma is the exponent power of the power law transformation mechanism, if gamma is more than 1, the darker original X-ray image can be darkened, namely, the area range of high gray values is amplified, and the amplification degree is higher than that of the low gray value range; the effect that the low gray scale range is compressed and the high gray scale range is enlarged is achieved, refer to fig. 3, and the graying ray image shows different effects under the condition that the gamma values are different.
After the gray-scale radiographic image is converted into a binary sequence, the binary sequence is expressed by a digital sequence formed by 0 or 1, and one gray-scale radiographic image is an 8-bit binary sequence, because the pixel value of the image is 0-255, h (t) can be used for expressing the binary sequence, and the gray-scale radiographic image comprises image information and noise information. Stochastic resonance processing is carried out on the binary sequence, and a preset FHN neuron parallel array model can be obtained:
dv i /dt＝v i (t)×(a-v i (t))×(v i (t)-1)-b/y×v i (t)+h(t)
wherein v is i And (t) is a voltage variable at the moment t, i=1, 2,3, … …, L and L are the array sizes of the preset FHN neuron parallel array model, and y, a and b are all basic parameters.
The processing of the gray-scale radiographic image by adopting the preset FHN neuron parallel array model is to perform denoising processing based on the FHN neuron parallel array of stochastic resonance, so that a good denoising effect can be achieved, and referring to fig. 4 and 5, fig. 4 and 5 are signal and corresponding spectrum contrast diagrams before and after denoising processing.
Further, step 101 further comprises:
the optimized radiographic image is subjected to preprocessing operations including edge detection, dilation, cleaning, segmentation and compression.
The preprocessing operation procedure is added in this embodiment to further improve the image quality, and to facilitate the subsequent image processing. The edge detection is to detect the key part of the ray image, and because the edge of the ray image is obvious and a large amount of noise can be filtered through denoising treatment, the embodiment selects an algorithm with high calculation speed to perform the edge detection, namely a Robert edge detection algorithm.
When the edge detection is performed on the radial image, the situation that the critical part is missing is unavoidable, so that the missing part is filled by adopting expansion processing in the embodiment, and the critical part is completely filled. The image cleaning is mainly to remove non-critical parts of the expanded radiographic image, such as saw teeth, white spots or parts of the edge of the image extending outwards from the boundary of the critical part by setting a cleaning threshold; the boundary of the key part is preset with the highest gray value and the value 255. The segmentation is to cut the key parts of the radiation image obtained after cleaning to obtain key part information, and then to compress the size to obtain the radiation image of the target size. It will be appreciated that other preprocessing algorithms may be selected to process the image according to the actual situation, and the preprocessing method is not limited herein.
And 102, extracting the features of the optimized radiographic image by adopting an improved principal component analysis method to obtain a low-dimensional feature vector, wherein the improved principal component analysis method comprises a kernel function.
Further, step 102 includes:
after expressing the optimized ray image as an image data matrix, mapping the image data matrix to a high-dimensional feature space by improving a kernel function in a principal component analysis method to obtain a high-dimensional feature expression;
and carrying out principal component analysis calculation based on the feature vector according to the high-dimensional feature expression to obtain a low-dimensional feature vector.
It should be noted that, the conventional Principal Component Analysis (PCA) algorithm cannot process nonlinear data, so in this embodiment, a kernel function is used to improve the principal component analysis method, so as to obtain an improved principal component analysis method, which is used to extract features in an optimized radiographic image, so as to obtain a low-dimensional feature vector.
Specifically, the optimized radiographic image is expressed as an image data matrix x, and if there are M optimized radiographic images, the image data matrix is expressed as x k Wherein k=1, 2,3, once again, M, assuming:
where M is the total number of matrices, then its covariance matrix is expressed as:
matrix mapping of image data to low-dimensional feature space phi (x) using kernel functions k ) Assume that:
its corresponding covariance matrix is then:
for the followingAnd (3) carrying out feature vector analysis to enable the feature vector to be v, and expressing the feature equation as follows:
where λ is the eigenvalue. Based on the characteristic equation expression, the following can be obtained:
for the feature vector v, it can be expressed as:
wherein a is i As a coefficient matrix, the above formula is integrated to obtain:
wherein phi is i For a low-dimensional feature space, the projection result of the image data matrix in the high-dimensional feature space can be obtained through feature value and feature vector solving:
wherein K (x j X) is a reference to x j As a function of x,the correlation coefficient is a function and is a constant. The embodiment introduces the kernel function, but only adds the operation of the kernel function in the actual operation, so that the simple and efficient advantages of the PCA algorithm are reserved; and the operation process is nonlinear operation, so that the problem that the PCA algorithm cannot process nonlinear structure data is effectively solved.
And step 103, inputting the low-dimensional feature vector into an improved KNN model for classifying the crimping defects to obtain a crimping defect recognition result, wherein the improved KNN model is obtained by optimizing a particle swarm algorithm.
The improved KNN model is a target model which can be directly used for classifying the crimping defects after training, the crimping conditions in the image of the current research and analysis are analyzed through the model, and specific results, namely crimping defect recognition results, are classified. The particle swarm algorithm is mainly used for updating the number K of adjacent samples in the KNN model, plays a role in optimizing the KNN model, and obtains an improved KNN model.
Further, step 103, further includes:
classifying and training the initial KNN model based on the number of the initialized adjacent samples and a preset training data set, and calculating classification errors;
taking the classification error as a fitness value, and updating the initial number of adjacent samples to obtain the updated number of adjacent samples;
and returning to the step of classifying and training the initial KNN model based on the initialized adjacent sample number and the preset training data set based on the updated adjacent sample number until the iteration termination condition is reached, so as to obtain the optimal adjacent sample number and the improved KNN model.
It should be noted that, the KNN (K Nearest Neighbours, KNN) algorithm belongs to a supervised learning algorithm, that is, is a selection of K nearest samples, and when a new sample value is to be predicted, it is possible to determine which classifications the K sample values closest to the sample belong to, and determine the classification of the current sample value based on the classification.
Each particle in the particle swarm algorithm (Particle Swarm Optimization, PSO) can be regarded as a search individual in the N-dimensional search space, the current position of the particle is a candidate solution of the corresponding optimization problem, and the flight process of the particle is the search process of the individual; the flight speed of the particles can be dynamically adjusted according to the historical optimal position of the particles and the historical optimal position of the population. The particles have only two properties: speed, which represents the speed of movement, and position, which represents the direction of movement. The optimal solution searched by each particle is called an individual extremum, and the optimal individual extremum in the particle swarm is used as the current global optimal solution. Continuously iterating, and updating the speed and the position; and finally obtaining the optimal solution meeting the termination condition.
The adjacent distance of the KNN model in this embodiment is selected as the euclidean distance, and the two-dimensional space is calculated as follows:
the multi-dimensional space is calculated as follows:
wherein x is i 、y i The coordinate point of the sample i is given, and n is the total sample number. The specific category condition of the Y to be classified can be judged by calculating Euclidean distance between the Y to be classified and all the points and then selecting the classification condition of the first K points through storage and sorting.
The construction process of the improved KNN model in the embodiment specifically comprises the following steps: firstly initializing the number K of adjacent samples, then carrying out classification training on an initial KNN model based on a large number of preset training data sets to obtain a prediction result, and calculating a classification error according to the prediction result and an actual class; then taking the classification error as a fitness value, starting particle speed and position updating based on a particle swarm optimization algorithm, namely updating the initial number of adjacent samples to obtain the updated number of adjacent samples; and then returning to the step of classifying and training the initial KNN model based on the number of the initialized adjacent samples and the preset training data set, and iterating continuously until the iteration termination condition is reached, so that the optimal number of the adjacent samples and the corresponding improved KNN model can be obtained, and the model can be directly used in a scene of identifying the actual crimping defects. It is to be understood that the iteration termination condition may be a preset maximum iteration number, or may be other iteration conditions, which is not limited herein.
The transmission line crimping defect identification method based on the X-ray image provided by the embodiment of the application adopts an improved graying algorithm and a denoising processing method to carry out graying processing and denoising processing on an original X-ray image, and is used for improving the quality of the original X-ray image, reducing the pressure of subsequent image processing and improving the accuracy of defect identification from the source; in addition, in order to overcome the defect that the principal component analysis method lacks of nonlinear expression capability, a kernel function is introduced to improve the principal component analysis method, so that the nonlinear expression capability of the extracted low-dimensional feature vector is stronger; moreover, the improved KNN model optimized by the particle swarm optimization can further ensure the reliability of the crimping defect identification result. Therefore, the embodiment of the application can solve the technical problems that the prior art cannot overcome the quality defect of the X-ray image and ensure the accuracy of the identification result of the image defect, thereby failing to meet the actual application requirements.
For ease of understanding, referring to fig. 2, the present application provides an embodiment of a transmission line crimping defect identification device based on an X-ray image, including:
the image processing unit 201 is configured to respectively perform graying processing and denoising processing on an original X-ray image of a strain clamp of a power transmission line based on a preset adaptive graying algorithm and a preset FHN neuron parallel array model, so as to obtain an optimized ray image, where the preset adaptive graying algorithm includes a power law transformation mechanism, and the preset FHN neuron parallel array model is constructed based on a stochastic resonance mechanism;
a feature extraction unit 202, configured to extract features of the optimized radiographic image by using an improved principal component analysis method, to obtain a low-dimensional feature vector, where the improved principal component analysis method includes a kernel function;
the defect recognition unit 203 is configured to input the low-dimensional feature vector into an improved KNN model for performing crimp defect classification, so as to obtain a crimp defect recognition result, where the improved KNN model is optimized by a particle swarm algorithm.
Further, the image processing unit 201 is specifically configured to:
constructing a power exponent transformation formula corresponding to a preset self-adaptive graying algorithm based on a power law transformation mechanism;
performing self-adaptive graying treatment on an original X-ray image of the power transmission line through a power exponent transformation formula to obtain a graying ray image;
after constructing a preset FHN neuron parallel array model based on a stochastic resonance mechanism, converting the gray-scale ray image into a binary sequence;
and denoising the binary sequence by adopting a preset FHN neuron parallel array model to obtain an optimized radiographic image.
Further, the method further comprises the following steps:
a preprocessing unit 204 for performing preprocessing operations on the optimized radiographic image, the preprocessing operations including edge detection, inflation, cleaning, segmentation, and compression.
Further, the feature extraction unit 202 is specifically configured to:
after expressing the optimized ray image as an image data matrix, mapping the image data matrix to a high-dimensional feature space by improving a kernel function in a principal component analysis method to obtain a high-dimensional feature expression;
and carrying out principal component analysis calculation based on the feature vector according to the high-dimensional feature expression to obtain a low-dimensional feature vector.
Further, the method further comprises the following steps:
a classification training unit 205, configured to perform classification training on the initial KNN model based on the initialized number of neighboring samples and a preset training data set, and calculate a classification error;
a parameter updating unit 206, configured to take the classification error as a fitness value, and update the initial number of neighboring samples to obtain an updated number of neighboring samples;
the iteration optimization unit 207 is configured to trigger the classification training unit based on the updated number of neighboring samples until an iteration termination condition is reached, thereby obtaining an optimal number of neighboring samples and an improved KNN model.
In the several embodiments provided by the present application, it should be understood that the disclosed apparatus and method may be implemented in other manners. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of the units is merely a logical function division, and there may be additional divisions when actually implemented, e.g., multiple units or components may be combined or integrated into another system, or some features may be omitted or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or units, which may be in electrical, mechanical or other form.
The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in the embodiments of the present application may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be embodied in essence or a part contributing to the prior art or all or part of the technical solution in the form of a software product stored in a storage medium, comprising several instructions for executing all or part of the steps of the method according to the embodiments of the present application by means of a computer device (which may be a personal computer, a server, or a network device, etc.). And the aforementioned storage medium includes: u disk, mobile hard disk, read-Only Memory (ROM), random access Memory (Random Access Memory, RAM), magnetic disk or optical disk, etc.
The above embodiments are only for illustrating the technical solution of the present application, and not for limiting the same; although the application has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present application.
Claims (10)
1. The transmission line crimping defect identification method based on the X-ray image is characterized by comprising the following steps of:
respectively carrying out graying treatment and denoising treatment on an original X-ray image of a strain clamp of a power transmission line based on a preset self-adaptive graying algorithm and a preset FHN neuron parallel array model to obtain an optimized ray image, wherein the preset self-adaptive graying algorithm comprises a power law transformation mechanism, and the preset FHN neuron parallel array model is constructed based on a stochastic resonance mechanism;
extracting the characteristics of the optimized radiographic image by adopting an improved principal component analysis method to obtain a low-dimensional characteristic vector, wherein the improved principal component analysis method comprises a kernel function;
and inputting the low-dimensional feature vector into an improved KNN model for compression joint defect classification to obtain compression joint defect identification results, wherein the improved KNN model is obtained by optimizing a particle swarm algorithm.
2. The method for identifying the crimping defect of the transmission line based on the X-ray image according to claim 1, wherein the method for respectively performing the graying treatment and the denoising treatment on the original X-ray image of the strain clamp of the transmission line based on the preset adaptive graying algorithm and the preset FHN neuron parallel array model to obtain the optimized ray image comprises the following steps:
constructing a power exponent transformation formula corresponding to a preset self-adaptive graying algorithm based on a power law transformation mechanism;
performing self-adaptive graying treatment on an original X-ray image of the power transmission line through the power exponent transformation formula to obtain a graying ray image;
after a preset FHN neuron parallel array model is established based on a stochastic resonance mechanism, converting the gray-scale ray image into a binary sequence;
and denoising the binary sequence by adopting the preset FHN neuron parallel array model to obtain an optimized radiographic image.
3. The method for identifying the crimping defect of the power transmission line based on the X-ray image according to claim 1, wherein the method is characterized in that the method comprises the steps of respectively carrying out graying treatment and denoising treatment on an original X-ray image of a strain clamp of the power transmission line based on a preset self-adaptive graying algorithm and a preset FHN neuron parallel array model to obtain an optimized ray image, and further comprises the following steps:
the optimized radiographic image is subjected to preprocessing operations including edge detection, dilation, sweeping, segmentation and compression.
4. The method for identifying a crimp defect in an X-ray image based power line according to claim 1, wherein the method for extracting features of the optimized ray image by using an improved principal component analysis method to obtain a low-dimensional feature vector, the improved principal component analysis method comprising a kernel function comprises:
after the optimized radiographic image is expressed as an image data matrix, mapping the image data matrix to a high-dimensional feature space by improving a kernel function in a principal component analysis method to obtain a high-dimensional feature expression;
and carrying out principal component analysis and calculation based on the feature vector according to the high-dimensional feature expression to obtain a low-dimensional feature vector.
5. The method for identifying the crimping defects of the power transmission line based on the X-ray image according to claim 1, wherein the step of inputting the low-dimensional feature vector into an improved KNN model for crimping defect classification to obtain a crimping defect identification result, the improved KNN model is obtained by optimizing a particle swarm algorithm, and the method further comprises the following steps:
classifying and training the initial KNN model based on the number of the initialized adjacent samples and a preset training data set, and calculating classification errors;
taking the classification error as a fitness value, and updating the initial adjacent sample number to obtain an updated adjacent sample number;
and returning the step of classifying and training the initial KNN model based on the initialized adjacent sample number and the preset training data set based on the updated adjacent sample number until the iteration termination condition is reached, so as to obtain the optimal adjacent sample number and the improved KNN model.
6. X-ray image-based transmission line crimping defect identification device, which is characterized by comprising:
the image processing unit is used for respectively carrying out graying treatment and denoising treatment on an original X-ray image of a strain clamp of the power transmission line based on a preset self-adaptive graying algorithm and a preset FHN neuron parallel array model to obtain an optimized ray image, wherein the preset self-adaptive graying algorithm comprises a power law transformation mechanism, and the preset FHN neuron parallel array model is constructed based on a stochastic resonance mechanism;
the feature extraction unit is used for extracting the features of the optimized radiographic image by adopting an improved principal component analysis method to obtain a low-dimensional feature vector, wherein the improved principal component analysis method comprises a kernel function;
and the defect identification unit is used for inputting the low-dimensional feature vector into an improved KNN model for compression joint defect classification to obtain compression joint defect identification results, and the improved KNN model is obtained by optimizing a particle swarm algorithm.
7. The X-ray image based power line crimping defect identification device in accordance with claim 6, wherein the image processing unit is specifically configured to:
constructing a power exponent transformation formula corresponding to a preset self-adaptive graying algorithm based on a power law transformation mechanism;
performing self-adaptive graying treatment on an original X-ray image of the power transmission line through the power exponent transformation formula to obtain a graying ray image;
after a preset FHN neuron parallel array model is established based on a stochastic resonance mechanism, converting the gray-scale ray image into a binary sequence;
and denoising the binary sequence by adopting the preset FHN neuron parallel array model to obtain an optimized radiographic image.
8. The X-ray image based power line crimping defect identification device in accordance with claim 6, further comprising:
and the preprocessing unit is used for preprocessing the optimized radiographic image, and the preprocessing operation comprises edge detection, expansion, cleaning, segmentation and compression.
9. The X-ray image based power line crimping defect identification device in accordance with claim 6, wherein the feature extraction unit is specifically configured to:
after the optimized radiographic image is expressed as an image data matrix, mapping the image data matrix to a high-dimensional feature space by improving a kernel function in a principal component analysis method to obtain a high-dimensional feature expression;
and carrying out principal component analysis and calculation based on the feature vector according to the high-dimensional feature expression to obtain a low-dimensional feature vector.
10. The X-ray image based power line crimping defect identification device in accordance with claim 6, further comprising:
the classification training unit is used for carrying out classification training on the initial KNN model based on the number of the initialized adjacent samples and a preset training data set, and calculating classification errors;
a parameter updating unit, configured to take the classification error as a fitness value, and update the initial number of neighboring samples to obtain an updated number of neighboring samples;
and the iterative optimization unit is used for triggering the classification training unit based on the updated adjacent sample number until an iterative termination condition is reached, so as to obtain the optimal adjacent sample number and an improved KNN model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310948046.5A CN116977853A (en) | 2023-07-28 | 2023-07-28 | X-ray image-based transmission line crimping defect identification method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310948046.5A CN116977853A (en) | 2023-07-28 | 2023-07-28 | X-ray image-based transmission line crimping defect identification method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116977853A true CN116977853A (en) | 2023-10-31 |
Family
ID=88474456
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310948046.5A Pending CN116977853A (en) | 2023-07-28 | 2023-07-28 | X-ray image-based transmission line crimping defect identification method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116977853A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117876330A (en) * | 2024-01-11 | 2024-04-12 | 广东粤电科试验检测技术有限公司 | Strain clamp crimping defect detection method and system based on X-ray image identification |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102693529A (en) * | 2012-05-24 | 2012-09-26 | 杭州电子科技大学 | Grayscale image enhancement method based on stochastic resonance mechanism of delayed self-feedback FHN (fitzhugh-nagumo) model |
| CN102915528A (en) * | 2012-11-06 | 2013-02-06 | 杭州电子科技大学 | Method for enhancing binary image of array cascade FHN (FitzHugh Nagumo) model stochastic resonance mechanism |
| CN109164343A (en) * | 2018-08-30 | 2019-01-08 | 西华大学 | Quantify the Diagnosis Method of Transformer Faults with weighting KNN based on characteristic information |
| CN111402208A (en) * | 2020-03-03 | 2020-07-10 | 天津理工大学 | Incremental-based weld surface defect feature extraction and classification identification method |
-
2023
- 2023-07-28 CN CN202310948046.5A patent/CN116977853A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102693529A (en) * | 2012-05-24 | 2012-09-26 | 杭州电子科技大学 | Grayscale image enhancement method based on stochastic resonance mechanism of delayed self-feedback FHN (fitzhugh-nagumo) model |
| CN102915528A (en) * | 2012-11-06 | 2013-02-06 | 杭州电子科技大学 | Method for enhancing binary image of array cascade FHN (FitzHugh Nagumo) model stochastic resonance mechanism |
| CN109164343A (en) * | 2018-08-30 | 2019-01-08 | 西华大学 | Quantify the Diagnosis Method of Transformer Faults with weighting KNN based on characteristic information |
| CN111402208A (en) * | 2020-03-03 | 2020-07-10 | 天津理工大学 | Incremental-based weld surface defect feature extraction and classification identification method |
Non-Patent Citations (4)
| Title |
|---|
| 任江涛;卓晓岚;许盛灿;印鉴;: "基于PSO面向K近邻分类的特征权重学习算法", 计算机科学, no. 05, 25 May 2007 (2007-05-25), pages 4 * |
| 任江涛;姚树宇;纪庆革;: "一种基于PSO同步进行特征选择及参数优化的核K近邻分类算法", 小型微型计算机系统, no. 08, 15 August 2007 (2007-08-15) * |
| 孙洪央;徐祖洋;王静;雷沛;吴开杰;柴新禹;: "基于PSO-kNN算法与多生理参数的压力状态下情绪识别", 中国医疗器械杂志, no. 02, 30 March 2013 (2013-03-30) * |
| 李欢;焦建民;: "简化的粒子群优化快速KNN分类算法", 计算机工程与应用, no. 32, 11 November 2008 (2008-11-11) * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117876330A (en) * | 2024-01-11 | 2024-04-12 | 广东粤电科试验检测技术有限公司 | Strain clamp crimping defect detection method and system based on X-ray image identification |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109977943B (en) | Image target recognition method, system and storage medium based on YOLO | |
| CN107908963B (en) | Method for automatically detecting core characteristics of malicious codes | |
| EP3171332B1 (en) | Methods and systems for inspecting goods | |
| CN111524137B (en) | Cell identification counting method and device based on image identification and computer equipment | |
| JP7294695B2 (en) | Program, Information Recording Medium, Classification Apparatus, and Classification Method Based on Trained Model | |
| CN108520215B (en) | Single-sample face recognition method based on multi-scale joint feature encoder | |
| CN111915572A (en) | Self-adaptive gear pitting quantitative detection system and method based on deep learning | |
| CN116910752B (en) | Malicious code detection method based on big data | |
| US20230169555A1 (en) | System and method for textual analysis of images | |
| CN111400528B (en) | Image compression method, device, server and storage medium | |
| CN116977853A (en) | X-ray image-based transmission line crimping defect identification method and device | |
| Akhtar et al. | Optical character recognition (OCR) using partial least square (PLS) based feature reduction: an application to artificial intelligence for biometric identification | |
| CN110766708A (en) | Image comparison method based on contour similarity | |
| CN112132151A (en) | Image character recognition system and method based on recurrent neural network recognition algorithm | |
| CN111241550B (en) | Vulnerability detection method based on binary mapping and deep learning | |
| CN114926680A (en) | Malicious software classification method and system based on AlexNet network model | |
| Velliangira et al. | A novel forgery detection in image frames of the videos using enhanced convolutional neural network in face images | |
| CN115861226A (en) | Method for intelligently identifying surface defects by using deep neural network based on characteristic value gradient change | |
| CN116206227B (en) | Picture examination system and method for 5G rich media information, electronic equipment and medium | |
| Hameed et al. | Content based image retrieval based on feature fusion and support vector machine | |
| CN114065798A (en) | Visual identification method and device based on machine identification | |
| CN111639212A (en) | Image retrieval method in mining intelligent video analysis | |
| CN115019305B (en) | Method, device and equipment for identifying root tip cells and readable storage medium | |
| US11715288B2 (en) | Optical character recognition using specialized confidence functions | |
| CN115292702A (en) | Malicious code family identification method, device, equipment and storage medium |
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 |