CN116229158A - Image classification labeling method based on artificial intelligence - Google Patents
Image classification labeling method based on artificial intelligence Download PDFInfo
- Publication number
- CN116229158A CN116229158A CN202310100731.2A CN202310100731A CN116229158A CN 116229158 A CN116229158 A CN 116229158A CN 202310100731 A CN202310100731 A CN 202310100731A CN 116229158 A CN116229158 A CN 116229158A
- Authority
- CN
- China
- Prior art keywords
- image
- wolf
- gray
- individual
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000002372 labelling Methods 0.000 title claims abstract description 22
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 12
- 241000282461 Canis lupus Species 0.000 claims abstract description 95
- 238000000034 method Methods 0.000 claims abstract description 30
- 238000012706 support-vector machine Methods 0.000 claims abstract description 25
- 238000007781 pre-processing Methods 0.000 claims abstract description 8
- 230000008569 process Effects 0.000 claims abstract description 7
- 238000012549 training Methods 0.000 claims abstract description 7
- 238000013145 classification model Methods 0.000 claims abstract description 4
- 238000012360 testing method Methods 0.000 claims abstract description 4
- 230000006870 function Effects 0.000 claims description 26
- 238000005457 optimization Methods 0.000 claims description 16
- 241000282421 Canidae Species 0.000 claims description 9
- 238000001914 filtration Methods 0.000 claims description 8
- 230000002146 bilateral effect Effects 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 4
- 238000012216 screening Methods 0.000 claims description 4
- 238000002790 cross-validation Methods 0.000 claims description 3
- 230000003247 decreasing effect Effects 0.000 claims description 3
- 230000002708 enhancing effect Effects 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 230000008859 change Effects 0.000 description 4
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000003709 image segmentation Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 208000003464 asthenopia Diseases 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000008094 contradictory effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
Classifications
-
- 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
- G06V10/765—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects using rules for classification or partitioning the feature space
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- 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
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- 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
- G06V10/776—Validation; Performance evaluation
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (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 invention provides an image classification labeling method based on artificial intelligence, which comprises the following steps: acquiring a marked image data set, and preprocessing the image data set, wherein the preprocessing process comprises denoising operation, image enhancement operation and the like; and constructing an autonomous classification model by adopting a support vector machine (Support Vector Machine, SVM) to a training image set, wherein the acquisition of optimal parameters in the support vector machine adopts an improved gray wolf algorithm, and the image set is used for testing the quality (obtaining a fitness value) of the trained SVM model. And carrying out autonomous classification labeling on the input unlabeled image based on the SVM model with the optimal parameters. The method effectively carries out autonomous classification labeling on the images, remarkably improves classification accuracy, has stronger anti-interference capability, and solves the problem that the existing images cannot be subjected to autonomous classification labeling.
Description
Technical Field
The invention relates to the technical field of image classification, in particular to an image classification labeling method based on artificial intelligence.
Background
With the development of the information age, there is a huge amount of image data in life, so in order to acquire useful image information therein, it is necessary to classify and annotate different types of images with different structures. Meanwhile, models such as machine learning, neural networks and the like also depend on classification labels of image data. However, the conventional image classification labeling technology is a time-consuming and labor-consuming process, and requires manual screening and labeling, so that not only is a lot of manual cost consumed, but also human errors are large, namely, human uncertainty factors such as human eye fatigue, distraction, false marks and the like exist, so that the image classification labeling is wrong. Meanwhile, the existing autonomous image classification labeling method needs a large amount of calculation cost, and has the problems of long calculation time, poor classification effect and the like.
There is therefore a need for an adaptive dermoscope image segmentation method based on artificial intelligence to solve the above problems.
Disclosure of Invention
The invention discloses an image classification labeling method based on artificial intelligence, which provides a self-adaptive, efficient and high-interference-resistance automatic image classification labeling method, and effectively improves the automatic image classification labeling, thereby effectively solving the technical problems in the background technology.
In order to achieve the above purpose, the technical scheme of the invention is as follows:
an artificial intelligence-based adaptive dermoscope image segmentation method comprises the following steps:
s1: acquiring a marked image data set, and preprocessing all images in the image data set, wherein the preprocessing process comprises denoising operation and image enhancement operation;
s2: the method comprises the steps of constructing an autonomous classification model for an image training set by adopting a support vector machine, wherein the acquisition of optimal parameters in the support vector machine adopts an improved gray wolf optimization algorithm, and testing the quality of the trained support vector machine model by using the image training set, wherein the method comprises the following steps:
s21: there are multiple first parameters in the support vector machine model, wherein the main parameters include penalty factors and kernel parameters, initializing various second parameters of the gray wolf population, the second parameters including the number N of the gray wolf population, the number D of the dimensions and the maximum number T of iterations max ；
S22: setting a boundary value of a first parameter in a support vector machine model, and obtaining an initialized wolf population X= { X according to the boundary value of the first parameter 1 ，X 2 ，...，X N X is a collection of gray wolf populations, X i (i=1, 2,., N) represents the individual of the ith wolf,
position information representing the ith gray wolf individual,/->
Position information representing the j-th dimension of the i-th wolf individual;
s23: calculating the corresponding fitness function of each position of the wolves to obtain fitness values, and screening out the previous three optimal fitness values F 1 ，F 2 ，F 3 Will adapt to the value F 1 ，F 2 ，F 3 The mapped three positions are denoted as X a ，X b ，X c Position X where the best fitness is mapped a Is the optimal solution;
s24: updating each parameter factor of the wolves and the position of each wolf, and judging whether the updated position of the wolf population exceeds a boundary value or not;
s25: if the fitness value of the updated wolf individual is larger than the fitness value before updating, replacing the position of the updated wolf individual with the position before updating; if the fitness value of the updated wolf individual is smaller than the fitness value before updating, the position of the wolf individual before updating is reserved; iterating for several times until reaching the maximum iteration times, and outputting the position information X of the gray wolf individual with the maximum fitness value a ；
S3: and the support vector machine model based on the optimal parameters carries out autonomous classification labeling on the input unlabeled image.
As a preferred modification of the present invention, step S1 includes:
s11: the denoising operation is carried out on the images in the image data by adopting a bilateral filtering method, and the specific expression is as follows:
wherein W is p Is a standard quantity:
wherein S represents a pixel point set of an image, q represents a pixel point, IM represents a denoised image, W p To normalize the parameters, sigma d Sum sigma r The filtering amount of the image I is measured and is respectively based on the spatial distance standard deviation and the gray level similarity standard deviation of the Gaussian function,
is a spatial function for reducing the effect of remote pixels; />
Is a range function for reducing gray values different from I p Is not affected by the pixel q. p is the template center pixel. I Q Is a noisy image, I p Is a template image;
typically, spatial proximity functions of bilateral filters
And gray level similarity->
A gaussian function with parameters euclidean distance is taken and is generally defined as:
wherein d (p, q) and delta (I (p), I (q)) are respectively the euclidean distance of the image volume pixel point and the gray level difference of the pixel; i (p) and I (q) represent the gray values of p and q, respectively;
s12: the method for enhancing the image by adopting the histogram equalization method comprises the following steps:
firstly counting the number of each gray value pixel in an image, secondly calculating the frequency of each gray value pixel, calculating the accumulated frequency, and finally mapping the image, wherein the gray value of the image is equal to the original gray value of the image.
As a preferable improvement of the invention, in step S23, the fitness function adopts the svmtrain function in the LIBSVM to carry out K-fold cross validation, so as to obtain the average classification accuracy.
As a preferred modification of the present invention, step S24 specifically includes:
s241: based on the first three optimal fitness values F 1 ，F 2 ，F 3 Mapped position information X a ，X b ，X c The position update of the individual wolves based on the wolf algorithm is as follows:
wherein X is i (t) and X i (t+1) is the position of the ith gray wolf individual at the moment of the current iteration time t and the moment of the next iteration time t+1, X a (t)，X b (t)，X c (t) represents the position mapped by the first three optimal fitness values until time t;
part of the parameters in the gray wolf algorithm are updated as follows:
A＝2a·r 1 -a
C＝2·r 2
where T is the T-th iteration, T max Is the maximum iteration number, a is linearly decreasing from 2 to 0 as the iteration number increases, r 1 And r 2 Is [0,1 ]]A=a, random vector of (a) 1 ，A 2 ，A 3 C=c 1 ，C 2 ，C 3 ；
The improved gray wolf optimization algorithm proposed this time is optimized from the following two aspects:
firstly, the value of the convergence factor a adopts normal distribution instead of linear change, so that the global optimization capacity of the early stage and the local optimization capacity of the later stage of the algorithm can be improved, and the convergence factor a is expressed by the following formula:
wherein sigma 1 Sum sigma 2 Is a variance parameter and satisfies
Setting sigma 1 ＝σ 2 ＝2；
Secondly, dividing the updating of the movement position of the wolf into two stages, and iterating the first half stage, namely 0 to
And the latter half of the iteration, i.e. +.>
To T max ；
The first half of the iteration stage should consider the wolf leader X a Is of importance, thus increasing the gray wolf leader X a The weight of (2) makes the individual a more prominent leader and the gray wolf position is updated as follows:
X i (t+1)＝w 1 X 1 +w 2 X 2 +w 3 X 3
w 1 ，w 2 ，w 3 three weight values, here set to w 1 ＝0.4，w 2 ＝0.3，w 3 ＝0.3；
In the latter half of the iteration stage, the wolf leader X a The position of (2) is not necessarily the best in the later wolf group, and the later wolf group is easy to sink into local optimum, so that the problem that an optimization result is sunk into local optimum is effectively avoided by adopting the weight setting of a sine and cosine function, and the method is expressed as follows:
whether the updated X (t+1) position of the improved wolf algorithm exceeds the parameter boundary value of the support vector machine model or not, and if the individual position is smaller than the minimum parameter boundary value, assigning a parameter minimum boundary value; if the individual position is greater than the maximum parameter boundary value, the maximum parameter boundary value is assigned.
The beneficial effects of the invention are as follows:
1. the double-sided filtering is adopted to carry out denoising operation on the image, so that noise interference in the image acquisition process is effectively removed, the image quality is improved, and the subsequent image processing operation is facilitated;
2. the histogram equalization is adopted to enhance the image, so that the visual effect of the image is enhanced, meanwhile, the original unclear image is changed into clear or the interested features are emphasized, and the uninteresting features are restrained, so that the image quality is improved, and the information quantity of the image is enriched;
3. the gray wolf algorithm is applied to the support vector machine model to find the optimal parameter combination, and the optimal parameter combination is obtained through the self-adaption of the gray wolf algorithm, so that the traditional manual adjustment is replaced, the human cost and the time cost are greatly reduced, and meanwhile, the accuracy of image classification is improved;
4. the value of the convergence factor a adopted in the gray wolf algorithm adopts normal distribution instead of linear change, so that the global optimization capacity of the early stage and the local optimization capacity of the later stage of the algorithm can be improved;
5. in the method, the updating of the moving position of the wolf is divided into two stages in the wolf algorithm, the updating of the position of the front half stage of the wolf accelerates the convergence speed of the algorithm, and the position of the rear half stage of the wolf improves the robustness and stability of the algorithm, so that the convergence speed and the convergence precision of the whole algorithm are improved.
Detailed Description
The technical solutions of the embodiments of the present invention will be clearly and completely described in the following in conjunction with the embodiments of the present invention, and it is obvious that the described embodiments 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.
It should be noted that all directional indications (such as up, down, left, right, front, and rear … …) in the embodiments of the present invention are merely used to explain the relative positional relationship between the components, the movement condition, etc. in a specific posture, and if the specific posture is changed, the directional indication is changed accordingly.
Furthermore, descriptions such as those referred to as "first," "second," and the like, are provided for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implying an order of magnitude of the indicated technical features in the present disclosure. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include at least one such feature. In the description of the present invention, the meaning of "plurality" means at least two, for example, two, three, etc., unless specifically defined otherwise.
In the present invention, unless specifically stated and limited otherwise, the terms "connected," "affixed," and the like are to be construed broadly, and for example, "affixed" may be a fixed connection, a removable connection, or an integral body; can be mechanically or electrically connected; either directly or indirectly, through intermediaries, or both, may be in communication with each other or in interaction with each other, unless expressly defined otherwise. The specific meaning of the above terms in the present invention can be understood by those of ordinary skill in the art according to the specific circumstances.
In addition, the technical solutions of the embodiments of the present invention may be combined with each other, but it is necessary to be based on the fact that those skilled in the art can implement the technical solutions, and when the technical solutions are contradictory or cannot be implemented, the combination of the technical solutions should be considered as not existing, and not falling within the scope of protection claimed by the present invention.
The invention provides an image classification labeling method based on artificial intelligence, which comprises the following steps:
s1: acquiring a marked image data set, and preprocessing all images in the image data set, wherein the preprocessing process comprises denoising operation and image enhancement operation;
specifically, step S1 includes:
s11: the denoising operation is carried out on the images in the image data by adopting a bilateral filtering method, and the specific expression is as follows:
wherein W is p Is a standard quantity:
wherein S represents a pixel point set of an image, q represents a pixel point, IM represents a denoised image, W p To normalize the parameters, sigma d Sum sigma r The filtering amount of the image I is measured and is respectively based on the spatial distance standard deviation and the gray level similarity standard deviation of the Gaussian function,
is a spatial function for reducing the effect of remote pixels; />
Is a range function for reducing gray values different from I p Is not affected by the pixel q. p is the template center pixel. I q Is a noisy image, I p Is a template image;
typically, spatial proximity functions of bilateral filters
And gray level similarity->
A gaussian function with parameters euclidean distance is taken and is generally defined as:
wherein d (p, q) and delta (I (p), I (q)) are respectively the euclidean distance of the image volume pixel point and the gray level difference of the pixel; i (p) and I (q) represent the gray values of p and q, respectively;
s12: the method for enhancing the image by adopting the histogram equalization method comprises the following steps:
firstly counting the number of each gray value pixel in an image, secondly calculating the frequency of each gray value pixel, calculating the accumulated frequency, and finally mapping the image, wherein the gray value of the image is equal to the original gray value of the image.
S2: the method comprises the steps of constructing an autonomous classification model for an image training set by adopting a support vector machine, wherein the acquisition of optimal parameters in the support vector machine adopts an improved gray wolf optimization algorithm, and testing the quality of the trained support vector machine model by using the image training set, wherein the method comprises the following steps:
s21: there are multiple first parameters in the support vector machine model, where the main first parameters include penalty factors and kernel parameters, initializing the sirius populationEach second parameter including the number N of the wolf population, the number D of the dimensions and the maximum number T of the iterations max ；
S22: setting a boundary value of a first parameter in a support vector machine model, and obtaining an initialized wolf population X= { X according to the boundary value of the first parameter 1 ，X 2 ，...，X N X is a collection of gray wolf populations, X i (i=1, 2,., N) represents the individual of the ith wolf,
position information representing the ith gray wolf individual,/->
Position information representing the j-th dimension of the i-th wolf individual;
s23: calculating the corresponding fitness function of each position of the wolves to obtain fitness values, and screening out the previous three optimal fitness values F 1 ，F 2 ，F 3 Will adapt to the value F 1 ，F 2 ，F 3 The mapped three positions are denoted as X a ，X b ，X c Position X where the best fitness is mapped a Is the optimal solution;
specifically, step S23 includes:
and carrying out K-fold cross validation on the fitness function by adopting a svmtrain function in the LIBSVM to obtain the average classification accuracy.
S24: updating each parameter factor of the wolves and the position of each wolf, and judging whether the updated position of the wolf population exceeds a boundary value or not;
specifically, step S24 includes:
s241: based on the first three optimal fitness values F 1 ，F 2 ，F 3 Mapped position information X a ，X b ，X c The position update of the individual wolves based on the wolf algorithm is as follows:
wherein X is i (t) and X i (t+1) is the position of the ith gray wolf individual at the moment of the current iteration time t and the moment of the next iteration time t+1, X a (t)，X b (t)，X c (t) represents the position mapped by the first three optimal fitness values until time t;
part of the parameters in the gray wolf algorithm are updated as follows:
A＝2a·r 1 -a
C＝2·r 2
where T is the T-th iteration, T max Is the maximum iteration number, a is linearly decreasing from 2 to 0 as the iteration number increases, r 1 And r 2 Is [0,1 ]]A=a, random vector of (a) 1 ，A 2 ，A 3 C=c 1 ，C 2 ，C 3 ；
The improved gray wolf optimization algorithm proposed this time is optimized from the following two aspects:
firstly, the value of the convergence factor a adopts normal distribution instead of linear change, so that the global optimization capacity of the early stage and the local optimization capacity of the later stage of the algorithm can be improved, and the convergence factor a is expressed by the following formula:
wherein sigma 1 Sum sigma 2 Is a variance parameter and satisfies
Setting sigma 1 ＝σ 2 ＝2；
Secondly, dividing the updating of the movement position of the wolf into two stages, and iterating the first half stage, namely 0 to
And the latter half of the iteration, i.e. +.>
To T max ；
The first half of the iteration stage should consider the wolf leader X a Is of importance, thus increasing the gray wolf leader X a The weight of (2) makes the individual a more prominent leader and the gray wolf position is updated as follows:
X i (t+1)＝w 1 X 1 +w 2 X 2 +w 3 X 3
w 1 ，w 2 ，w 3 three weight values, here set to w 1 ＝0.4，w 2 ＝0.3，w 3 ＝0.3；
In the latter half of the iteration stage, the wolf leader X a The position of (2) is not necessarily the best in the later wolf group, and the later wolf group is easy to sink into local optimum, so that the problem that an optimization result is sunk into local optimum is effectively avoided by adopting the weight setting of a sine and cosine function, and the method is expressed by the following formula:
whether the updated X (t+1) position of the improved wolf algorithm exceeds the parameter boundary value of the SVM model or not, if the individual position is smaller than the minimum parameter boundary value, assigning the parameter minimum boundary value; if the individual position is greater than the maximum parameter boundary value, the maximum parameter boundary value is assigned.
S25: if the fitness value of the updated wolf individual is larger than the fitness value before updating, replacing the position of the updated wolf individual with the position before updating; if the fitness value of the updated individual wolf is smaller than the fitness value before updating, the individual wolf before updating is reservedIs a position of (2); iterating for several times until reaching the maximum iteration times, and outputting the position information X of the gray wolf individual with the maximum fitness value a ；
S3: and the support vector machine model based on the optimal parameters carries out autonomous classification labeling on the input unlabeled image.
The beneficial effects of the invention are as follows:
1. the double-sided filtering is adopted to carry out denoising operation on the image, so that noise interference in the image acquisition process is effectively removed, the image quality is improved, and subsequent image processing operation is facilitated.
2. The histogram equalization is adopted to enhance the image, so that the visual effect of the image is enhanced, meanwhile, the original unclear image is changed into clear or the interested features are emphasized, the uninteresting features are restrained, the image quality is improved, and the information quantity of the image is enriched.
3. The method adopts the gray wolf algorithm to be applied to the support vector machine model to find the optimal parameter combination, and the optimal parameter combination is obtained through the gray wolf algorithm in a self-adaptive way, so that the traditional manual adjustment is replaced, the human and time costs are greatly reduced, and the accuracy of image classification is improved.
4. The convergence factor a is adopted in the gray wolf algorithm, normal distribution is adopted instead of linear change, and the global optimization capacity of the early stage and the local optimization capacity of the later stage of the algorithm can be improved.
5. In the method, the moving position of the wolf is updated in two stages in the wolf algorithm, the front half stage of the wolf is updated so as to accelerate the convergence rate of the algorithm, and the rear half stage of the wolf is updated so as to improve the robustness and stability of the algorithm, so that the convergence rate and the convergence precision of the whole algorithm are improved;
although embodiments of the present invention have been disclosed above, it is not limited to the details of the description and the embodiments, which are well suited to various fields of use, additional modifications may be readily made by those skilled in the art without departing from the general concept defined by the claims and their equivalents.
Claims (4)
1. An image classification labeling method based on artificial intelligence is characterized by comprising the following steps:
s1: acquiring a marked image data set, and preprocessing all images in the image data set, wherein the preprocessing process comprises denoising operation and image enhancement operation;
s2: the method comprises the steps of constructing an autonomous classification model for an image training set by adopting a support vector machine, wherein the acquisition of optimal parameters in the support vector machine adopts an improved gray wolf optimization algorithm, and testing the quality of the trained support vector machine model by using the image training set, wherein the method comprises the following steps:
s21: there are multiple first parameters in the support vector machine model, wherein the first parameters comprise penalty factors and kernel parameters, and each second parameter of the gray wolf population is initialized, and the second parameters comprise the number N of the gray wolf population, the number D of the dimensions and the maximum iteration number T max ；
S22: setting a boundary value of a first parameter in a support vector machine model, and obtaining an initialized wolf population X= { X according to the boundary value of the first parameter 1 ，X 2 ，...，X N X is a collection of gray wolf populations, X i (i=1, 2,., N) represents the individual of the ith wolf,
position information representing the ith gray wolf individual,/->
Position information representing the j-th dimension of the i-th wolf individual;
s23: calculating the corresponding fitness function of each position of the wolves to obtain fitness values, and screening out the previous three optimal fitness values F 1 ，F 2 ，F 3 Will moderate value F 1 ，F 2 ，F 3 The mapped three positions are denoted as X a ，X b ，X c Of which the mostPosition X mapped by the fitness a Is the optimal solution;
s24: updating each parameter factor of the wolves and the position of each wolf, and judging whether the updated position of the wolf population exceeds a boundary value or not;
s25: if the fitness value of the updated wolf individual is larger than the fitness value before updating, replacing the position of the updated wolf individual with the position before updating; if the fitness value of the updated wolf individual is smaller than the fitness value before updating, the position of the wolf individual before updating is reserved; iterating for several times until reaching the maximum iteration times, and outputting the position information X of the gray wolf individual with the maximum fitness value a ；
S3: and the support vector machine model based on the optimal parameters carries out autonomous classification labeling on the input unlabeled image.
2. The image classification labeling method based on artificial intelligence according to claim 1, wherein step S1 comprises:
s11: the denoising operation is carried out on the images in the image data by adopting a bilateral filtering method, and the specific expression is as follows:
wherein W is p Is a standard quantity:
wherein S represents a pixel point set of an image, q represents a pixel point, IM represents a denoised image, W p To normalize the parameters, sigma d Sum sigma r The filtering amount of the image I is measured and is respectively based on the spatial distance standard deviation and the gray level similarity standard deviation of the Gaussian function,
is a space functionA number for reducing a long-range pixel effect; />
Is a range function for reducing gray values different from I p Is not affected by the pixel q. p is the template center pixel. I q Is a noisy image, I p Is a template image;
spatial proximity function of bilateral filter
And gray level similarity->
A gaussian function with parameters euclidean distance is taken and defined as: />
Wherein d (p, q) and delta (I (p), I (q)) are respectively the euclidean distance of the image volume pixel point and the gray level difference of the pixel; i (p) and I (q) represent the gray values of p and q, respectively;
s12: the method for enhancing the image by adopting the histogram equalization method comprises the following steps:
firstly counting the number of each gray value pixel in an image, secondly calculating the frequency of each gray value pixel, calculating the accumulated frequency, and finally mapping the image, wherein the gray value of the image is equal to the original gray value of the image.
3. The image classification labeling method based on artificial intelligence according to claim 1, wherein in step S23, the fitness function performs K-fold cross-validation by using svmtrain function in libvm to obtain average classification accuracy.
4. The image classification labeling method based on artificial intelligence according to claim 1, wherein in step S24, the improved gray wolf position update of the gray wolf algorithm and the update of part of parameters in the gray wolf algorithm specifically include the following steps:
s241: based on the first three optimal fitness values F 1 ，F 2 ，F 3 Mapped position information X a ，X b ，X c The position update of the individual wolves based on the wolf algorithm is as follows:
wherein X is i (t) and X i (t+1) is the position of the ith gray wolf individual at the moment of the current iteration time t and the moment of the next iteration time t+1, X a (t)，X b (t)，X c (t) represents the position mapped by the first three optimal fitness values until time t;
part of the parameters in the gray wolf algorithm are updated as follows:
A＝2a·r 1 -a
C＝2·r 2
where T is the T-th iteration, T max Is the maximum iteration number, a is linearly decreasing from 2 to 0 as the iteration number increases, r 1 And r 2 Is [0,1 ]]A=a, random vector of (a) 1 ，A 2 ，A 3 C=c 1 ，C 2 ，C 3 ；
The improved gray wolf optimization algorithm is optimized from two aspects:
firstly, the convergence factor a adopts normal distribution, and is expressed by the following formula:
wherein sigma 1 Sum sigma 2 Is a variance parameter and satisfies
Setting sigma 1 ＝σ 2 ＝2；
Secondly, dividing the updating of the movement position of the wolf into two stages, and iterating the first half stage, namely 0 to
And the latter half of the iteration, i.e. +.>
To T max ；
The first half of the iteration stage considers the gray wolf leader X a Is of importance, thus increasing the gray wolf leader X a The weight of (2) makes the individual a more prominent leader and the gray wolf position is updated as follows:
X i (t+1)＝w 1 X 1 +w 2 X 2 +w 3 X 3
w 1 ，w 2 ，w 3 is three weight values, set as w 1 ＝0.4，w 2 ＝0.3，w 3 ＝0.3；
The weight setting of the sine and cosine functions is adopted and expressed by the following formula:
whether the updated X (t+1) position of the improved wolf algorithm exceeds the parameter boundary value of the support vector machine model or not, and if the individual position is smaller than the minimum parameter boundary value, assigning a parameter minimum boundary value; if the individual position is greater than the maximum parameter boundary value, the maximum parameter boundary value is assigned.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310100731.2A CN116229158B (en) | 2023-02-08 | 2023-02-08 | Image classification labeling method based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310100731.2A CN116229158B (en) | 2023-02-08 | 2023-02-08 | Image classification labeling method based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116229158A true CN116229158A (en) | 2023-06-06 |
| CN116229158B CN116229158B (en) | 2023-09-05 |
Family
ID=86585202
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310100731.2A Active CN116229158B (en) | 2023-02-08 | 2023-02-08 | Image classification labeling method based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116229158B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104112143A (en) * | 2014-07-23 | 2014-10-22 | 大连民族学院 | Weighted hyper-sphere support vector machine algorithm based image classification method |
| US20150262083A1 (en) * | 2014-03-11 | 2015-09-17 | Siemens Aktiengesellschaft | Proximal Gradient Method for Huberized Support Vector Machine |
| CN111967670A (en) * | 2020-08-18 | 2020-11-20 | 浙江中新电力工程建设有限公司 | Switch cabinet partial discharge data identification method based on improved wolf algorithm |
| CN112085059A (en) * | 2020-08-06 | 2020-12-15 | 温州大学 | Breast cancer image feature selection method based on improved sine and cosine optimization algorithm |
| WO2021218120A1 (en) * | 2020-04-27 | 2021-11-04 | 江苏科技大学 | Method for fault identification of ship power device |
| CN115147445A (en) * | 2022-07-21 | 2022-10-04 | 潘奕旻 | Multi-threshold human brain x-ray image segmentation method based on mixed wolf algorithm |
| CN115392286A (en) * | 2022-07-20 | 2022-11-25 | 昆明理工大学 | Bearing fault diagnosis method based on 1dCNN-GWO-SVM model |
-
2023
- 2023-02-08 CN CN202310100731.2A patent/CN116229158B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150262083A1 (en) * | 2014-03-11 | 2015-09-17 | Siemens Aktiengesellschaft | Proximal Gradient Method for Huberized Support Vector Machine |
| CN104112143A (en) * | 2014-07-23 | 2014-10-22 | 大连民族学院 | Weighted hyper-sphere support vector machine algorithm based image classification method |
| WO2021218120A1 (en) * | 2020-04-27 | 2021-11-04 | 江苏科技大学 | Method for fault identification of ship power device |
| CN112085059A (en) * | 2020-08-06 | 2020-12-15 | 温州大学 | Breast cancer image feature selection method based on improved sine and cosine optimization algorithm |
| CN111967670A (en) * | 2020-08-18 | 2020-11-20 | 浙江中新电力工程建设有限公司 | Switch cabinet partial discharge data identification method based on improved wolf algorithm |
| CN115392286A (en) * | 2022-07-20 | 2022-11-25 | 昆明理工大学 | Bearing fault diagnosis method based on 1dCNN-GWO-SVM model |
| CN115147445A (en) * | 2022-07-21 | 2022-10-04 | 潘奕旻 | Multi-threshold human brain x-ray image segmentation method based on mixed wolf algorithm |
Non-Patent Citations (2)
| Title |
|---|
| YAQING FENG.ET.: "Point Cloud Registration Algorithm Based on the Grey Wolf Optimizer", 《IEEE ACCESS》, vol. 8, pages 143375 - 143382, XP011804719, DOI: 10.1109/ACCESS.2020.3013706 * |
| 徐辰华等: "基于正弦因子和量子局部搜索的灰狼优化算法", 《计算机工程与应用》, vol. 57, no. 24, pages 83 - 89 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116229158B (en) | 2023-09-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109389608B (en) | There is the fuzzy clustering image partition method of noise immunity using plane as cluster centre | |
| CN108986140B (en) | Target scale self-adaptive tracking method based on correlation filtering and color detection | |
| CN110211126B (en) | Image segmentation method based on intuitive fuzzy C-means clustering | |
| CN108664976B (en) | Super-pixel-based fuzzy spectral clustering brain tumor image automatic segmentation method | |
| CN115661467B (en) | Cerebrovascular image segmentation method, device, electronic equipment and storage medium | |
| CN103761726B (en) | Block adaptive image partition method based on FCM | |
| CN110853070A (en) | Underwater sea cucumber image segmentation method based on significance and Grabcut | |
| CN115908371B (en) | Plant leaf disease and pest degree detection method based on optimized segmentation | |
| CN106600610B (en) | FCM image segmentation method and device | |
| CN112750106A (en) | Nuclear staining cell counting method based on incomplete marker deep learning, computer equipment and storage medium | |
| CN110889865A (en) | Video target tracking method based on local weighted sparse feature selection | |
| CN111863232A (en) | Remote disease intelligent diagnosis system based on block chain and medical image | |
| CN107292896A (en) | Contour extraction method based on Snake models | |
| CN111582198A (en) | Automatic sea-land segmentation method for remote sensing image | |
| CN110163825B (en) | Human embryo heart ultrasonic image denoising and enhancing method | |
| CN108846845B (en) | SAR image segmentation method based on thumbnail and hierarchical fuzzy clustering | |
| CN113409335B (en) | Image segmentation method based on strong and weak joint semi-supervised intuitive fuzzy clustering | |
| CN116935133B (en) | Cardiovascular disease classification method and system based on SPECT image recognition | |
| CN116229158B (en) | Image classification labeling method based on artificial intelligence | |
| CN110826563B (en) | Finger vein segmentation method and device based on neural network and probability map model | |
| CN111127532B (en) | Medical image deformation registration method and system based on deep learning characteristic optical flow | |
| CN107169533B (en) | SAR image coastline detection algorithm of probability factor TMF of super-pixel | |
| CN116342653A (en) | Target tracking method, system, equipment and medium based on correlation filter | |
| CN115147445A (en) | Multi-threshold human brain x-ray image segmentation method based on mixed wolf algorithm | |
| CN110910392B (en) | Image segmentation method based on fuzzy clustering of multiple spatial information weighting |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: A Method of Image Classification and Annotation Based on Artificial Intelligence Effective date of registration: 20231026 Granted publication date: 20230905 Pledgee: China Construction Bank Corporation Qiannan Prefecture Branch Pledgor: GUIZHOU CATALIHUA INFORMATION TECHNOLOGY Co.,Ltd. Registration number: Y2023520000059 |
|
| PE01 | Entry into force of the registration of the contract for pledge of patent right |