CN112767432B - Nuclear intuition fuzzy clustering image segmentation method based on differential mutation grayish wolf optimization - Google Patents
Nuclear intuition fuzzy clustering image segmentation method based on differential mutation grayish wolf optimization Download PDFInfo
- Publication number
- CN112767432B CN112767432B CN202110242352.8A CN202110242352A CN112767432B CN 112767432 B CN112767432 B CN 112767432B CN 202110242352 A CN202110242352 A CN 202110242352A CN 112767432 B CN112767432 B CN 112767432B
- Authority
- CN
- China
- Prior art keywords
- wolf
- pixel
- gray
- spatial information
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The kernel intuitive fuzzy clustering image segmentation method based on the differential variation grayish wolf optimization comprises the following steps; step 1: setting the maximum iteration times, the clustering number and the Gaussian kernel function of the kernel intuitive fuzzy clustering algorithm; setting the size of a wolf colony and the maximum iteration times, and step 2: inputting an image; and step 3: extracting image space robust information; and 4, step 4: respectively calculating a membership matrix, a clustering center and a hesitation degree; and 5: judging the relation between a random variable rand1 and the maximum iteration times; and 6: calculating the fitness value of each gray wolf; and 7: executing a greedy mechanism, and updating alpha, beta, gamma wolf and position vectors thereof; and 8: if the algorithm termination condition is met, outputting the optimal gray wolf position; otherwise, returning to the step (4); and step 9: and (4) performing KIFCM segmentation on the image according to the optimal clustering center obtained in the step (8). The method is based on a dynamic random differential variation wolf pack position updating strategy, and optimization of a clustering center is achieved.
Description
Technical Field
The invention relates to the technical field of image segmentation, in particular to a kernel intuitive fuzzy clustering image segmentation method based on differential variation grayish wolf optimization.
Background
The traditional fuzzy clustering algorithm is easy to be trapped in local optimization during image segmentation, is sensitive to noise information, is sensitive to initial clustering centers and the like.
Disclosure of Invention
In order to solve the technical problems, the invention aims to provide a kernel intuitive fuzzy clustering image segmentation method based on differential variation grayish wolf optimization, and a kernel intuitive fuzzy clustering objective function fusing image space robust information is further constructed through an image space robust information extraction strategy; and a wolf group position updating strategy based on dynamic random differential variation is provided in the wolf optimization, so that the optimization of a clustering center is realized.
In order to achieve the purpose, the invention adopts the technical scheme that:
the kernel intuitive fuzzy clustering image segmentation method based on the differential variation grayish wolf optimization comprises the following steps;
step 1: setting the maximum iteration number of the kernel intuitive fuzzy clustering algorithm to be N max M =2, cluster number C, gaussian kernel function σ; setting the size of a wolf colony as N and the maximum iteration number as MaxDT;
step 2: input image X = { X = 1 ,x 2 ,…,x n }；
And step 3: extracting image space robust information;
and 4, step 4: respectively calculating membership degree matrix
Clustering center
And degree of hesitation
And 5: judging the relation between the random 1 and the maximum iteration time MaxDT;
step 6: according to the formula
Calculating the fitness value of each gray wolf;
and 7: according to the formula
Executing a greedy mechanism, and updating alpha, beta, gamma wolf and position vectors thereof; x i (t) is the current location of the gray wolf, X i ' (t) is the position of the gray wolf at the previous moment; fit (X) i (t)) is a fitness value at the current location of the gray wolf, fit (X' i (t)) is the fitness value of the position of the wolf at the previous moment;
and 8: if the algorithm end condition is met, outputting the optimal gray wolf position X α (i.e., the best cluster center sought); otherwise, returning to the step (4);
and step 9: and performing KIFCM segmentation on the image according to the optimal clustering center obtained in the step 8.
The third step is specifically as follows:
selecting a neighborhood window with the diameter of 5 to perform Z-shaped neighborhood window sliding without dead angles on the image so as to obtain new spatial information of image pixels,
is a neighborhood window with a pixel j as a center and a size of 5 multiplied by 5, and the gray value x of the pixel in the neighborhood window j The number of pixels of =0 is ζ, x j Number of pixels of =255 is η;
judging the relation between zeta + eta and the size of the domain window;
(1) When ζ + η =25, if ζ>Eta, then the spatial information of the updated pixel j is x jnew =0; otherwise if ζ<Eta, then the spatial information of the updated pixel j is x jnew ＝255；
(2) When zeta + eta is not equal to 25, all x in the neighborhood window are obtained j Not equal to 0 and x j Gray scale mean of not equal to 255 pixels
As spatial information for pixel j;
updating pixel spatial information, namely updating the spatial information of the pixel when the gray value of the pixel j and zeta and eta in a neighborhood window meet the following conditions;
①x j where ζ + η ≧ 2 and 0, the probability that the grayscale value of the pixel j is 0 is high, and the spatial information of the updated pixel j is:
②x j where =255 and ζ + η ≧ 2, the probability that the grayscale value of pixel j is 255 is large, and the spatial information of the updated pixel j is:
③x j ≠0，x j not equal to 255, and ζ + η is greater than or equal to 2, at this time, the probability that salt and pepper noise is hidden by other pixels in the neighborhood window is the largest, and the spatial information of the updated pixel j is as follows:
the specific algorithm of the step 4 is as follows:
the cluster number C, m =2 is a smoothing parameter, a gaussian kernel function K (x, y), and γ is a weighting factor of spatial information.
The determination method in the step 5 comprises the following steps:
(1) When rand1<1-t/MaxDT, (rand 1 is a random variable, the value of which varies between 0 and 1) depending on X i (t+1)＝X i1 (t)+φ(X i2 (t)-X i3 (t)) updating the gray wolf location, X i1 、X i2 、X i3 Is the position vector of any 3 grey wolfs in the wolf group, i1, i2, i3 are belonged to [1, N]And are not repeated with each other,
is a random scaling factor;
(2) When 1-t/MaxDT is not more than and rand1 is less than 1, according to
The position of the gray wolf is updated,
A＝(2×r 1 -1)×a、C＝2×r 2 a =2-2t/MaxDT, A, C is the coefficient vector, a is the linear convergence factor, linearly decreasing from 2-0 with the number of iterations, r 1 、r 2 Is [0, 1 ]]The random variable above, maxDT, is the maximum number of iterations.
The concrete formula of the step 6 is as follows:
J KIFCM (u, v) is represented by the formula
Obtaining; the cluster number C, m =2 is a smoothing parameter, a gaussian kernel function K (x, y), and γ is a weighting factor of spatial information.
The invention has the beneficial effects that:
the invention introduces dynamic random difference variation and a greedy mechanism to improve the wolf algorithm so as to further improve the convergence speed and precision of the algorithm; the method comprises the steps of suppressing the influence of noise information in image information by extracting image space robust information; segmentation experiments were performed on images containing gaussian noise, salt and pepper noise, and mixed noise. Simulation results show that the text algorithm has better segmentation efficiency and effect, can effectively overcome the influence of various noise information, and segments the image background and the target.
Drawings
FIG. 1 is a schematic flow chart of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings.
As shown in fig. 1: a kernel intuitive fuzzy clustering image segmentation method based on differential variation grayish wolf optimization comprises the following steps;
step 1: input image X = { X 1 ,x 2 ,…,x n }；
Step 2: and extracting image space robust information. And selecting a neighborhood window with the diameter of 5 to perform Z-shaped neighborhood window sliding without dead angles on the image, thereby obtaining new spatial information of image pixels.
Is a neighborhood window with a pixel j as a center and a size of 5 multiplied by 5, and the gray value x of the pixel in the neighborhood window j The number of pixels of =0 is ζ, x j The number of pixels of =255 is η. Determine ζ + η and fieldThe relationship of the size of the window is,
(1) When ζ + η =25, if ζ>Eta, then the spatial information of the updated pixel j is x jnew =0; otherwise if ζ<Eta, then the spatial information of the updated pixel j is x jnew ＝255；
(2) When zeta + eta is not equal to 25, all x in the neighborhood window are obtained j Not equal to 0 and x j Gray scale mean of not equal to 255 pixels
As the spatial information of the pixel j,
updating of pixel spatial information. And when the gray value of the pixel j and zeta and eta in the neighborhood window meet the following conditions, updating the spatial information of the pixel.
①x j ζ + η ≧ 2 and 0, the probability that the gray-scale value of the pixel j is 0 is high, and the spatial information of the updated pixel j is
②x j ζ + η ≧ 2 and =255, the probability that the gradation value of the pixel j is 255 at this time is high, and the spatial information of the update pixel j is
③x j ≠0，x j Not equal to 255, and zeta + eta is more than or equal to 2, at the moment, the possibility that other pixels in the neighborhood window imply salt and pepper noise is the largest, and the spatial information of the updated pixel j is
And step 3: setting kernel intuitive fuzzy clustering algorithmThe maximum number of iterations of the method is N max M =2, cluster number C, gaussian kernel function σ; setting the size of a wolf colony as N and the maximum iteration number as MaxDT;
and 4, step 4: according to the formula
And 5:
(1) When rand1<1-t/MaxDT, according to
X i (t+1)＝X i1 (t)+φ(X i2 (t)-X i3 (t)) updating the gray wolf location. X i1 、X i2 、X i3 Is the position vector of any 3 grey wolfs in the wolf group, i1, i2, i3 are belonged to [1, N]And are not repeated.
Is a random scaling factor.
(2) When 1-t/MaxDT is less than or equal to rand1 is less than 1, according to
The position of the grey wolf is updated,
a =2-2t/MaxDT, A, C is the coefficient vector, a is the linear convergence factor, and decreases linearly from 2-0 with the number of iterations. r is 1 、r 2 Is [0, 1 ]]A random variable of (c). MaxDT is the maximum number of iterations.
Step 6: according to the formula
Calculating the fitness value of each gray wolf;
and 7: according to the formula
Executing a greedy mechanism, and updating alpha, beta, gamma wolf and position vectors thereof;
and 8: if the algorithm end condition is met, outputting the optimal gray wolf position X α (i.e., the best cluster center sought); otherwise, returning to the step (4);
and step 9: and (4) performing KIFCM segmentation on the image according to the optimal clustering center obtained in the step (8).
In order to test the optimization performance of the improved gray wolf algorithm (marked as IGWO), six classical test functions f 1-f 6 shown in the table 1 are taken as objects, and the optimization result is compared with GWO algorithm and MGWO algorithm. Wherein: f 1-f 3 are variable dimension single mode functions, and f 4-f 6 are variable dimension multi-mode functions.
And (3) combining an improved grayish wolf optimization algorithm and a kernel intuitive fuzzy clustering algorithm, clustering on 6 functions and Iris data and segmenting the noise-containing image. The test result shows that the algorithm has good clustering effect and can realize better segmentation on the image containing the noise.
TABLE 1 classical test function
TABLE 2 comparison of optimized Performance for IGWO, MGWO and GWO
From Table 2, it can be seen that the function f is either a single mode function 1 ～f 3 Or a multi-modal function f 4 ～f 6 The optimizing precision of the IGWO algorithm is obviously higher than that of GWO and the MGWO algorithm.
In order to verify the kernel intuitive fuzzy clustering image segmentation method based on the differential variation grayish wolf optimization, the clustering number is selected to be 2, and the maximum iteration number is 100. And respectively adopting IFCM, KIFCM and IGWO _ KIFCM algorithms to segment images containing Gaussian noise, salt and pepper noise and mixed noise. According to the segmentation result, the IGWO _ KIFCM has the best segmentation effect on the image containing the noise, and meanwhile, the IGWO _ KIFCM optimizes the clustering center by adopting IGWO, so that the algorithm efficiency is greatly improved.
Claims (4)
1. The kernel intuitive fuzzy clustering image segmentation method based on the differential variation grayish wolf optimization is characterized by comprising the following steps of;
step 1: setting the maximum iteration number of the kernel intuitive fuzzy clustering algorithm to be N max M =2, cluster number C, gaussian kernel function σ; setting the size of a wolf group as N and the maximum iteration times as MaxDT;
step 2: input image X = { X 1 ,x 2 ,…,x n }；
And step 3: extracting image space robust information;
and 4, step 4: respectively calculating membership degree matrix
Clustering center
And degree of hesitation
And 5: judging the relation between a random variable rand1 and the maximum iteration number MaxDT;
step 6: according to the formula
Calculating the fitness value of each gray wolf; fit (u) is the fitness value at the current position of wolf's u, J KIFCM (u, v) is a target function of the KIFCM algorithm, u is any pixel in the image, and v is a clustering center;
and 7: according to the formula
Executing a greedy mechanism, and updating alpha, beta, gamma wolf and position vectors thereof; x i (t) is the current location of the gray wolf, X i ' (t) is the position of the gray wolf at the previous moment; fit (X) i (t)) is a fitness value at the current location of the gray wolf, fit (X' i (t)) is the fitness value of the position of the wolf at the previous moment;
and 8: if the algorithm end condition is met, outputting the optimal gray wolf position X a The obtained cluster center is the optimal cluster center; otherwise, returning to the step (4);
and step 9: performing KIFCM segmentation on the image according to the optimal clustering center obtained in the step 8;
the concrete formula of the step 6 is as follows:
J KIFCM (u, v) is represented by the formula
Obtaining; the cluster number C, m =2 is a smoothing parameter, a gaussian kernel function K (x, y), and γ is a weighting factor of spatial information.
2. The difference variant grayish wolf optimization-based kernel intuitive fuzzy clustering image segmentation method according to claim 1, wherein the step 3 is specifically:
selecting a neighborhood window with the diameter of 5 to perform Z-shaped neighborhood window sliding without dead angles on the image so as to obtain new spatial information of image pixels,
is a neighborhood window with a pixel j as a center and a size of 5 multiplied by 5, and the gray value x of the pixel in the neighborhood window j The number of pixels of =0 is ζ, x j Number of pixels of =255 is η;
judging the relation between zeta + eta and the size of the neighborhood window;
(1) When ζ + η =25, if ζ>Eta, then the spatial information of the updated pixel j is x jnew =0; otherwise if ζ<Eta, then the spatial information of the updated pixel j is x jnew ＝255；
(2) When zeta + eta ≠ 25, all x in the neighborhood window are obtained j Not equal to 0 and x j Gray scale mean of not equal to 255 pixels
As spatial information for pixel j;
updating pixel spatial information, namely updating the spatial information of the pixel when the gray value of the pixel j and zeta and eta in a neighborhood window meet the following conditions;
①x j =0, and ζ + η ≧ 2, the probability that the grayscale value of the pixel j is 0 at this time is high, and the spatial information of the update pixel j is:
②x j =255, and ζ + ηAnd 2, the probability that the gray value of the pixel j is 255 is high, and the spatial information of the updated pixel j is as follows:
③x j ≠0，x j and # 255, and ζ + η is greater than or equal to 2, at this time, the probability that salt and pepper noise is hidden by other pixels in the neighborhood window is the largest, and the spatial information of the updated pixel j is as follows:
3. the kernel-intuitive fuzzy clustering image segmentation method based on differential variation grayish wolf optimization according to claim 1, characterized in that the specific algorithm of the step 4 is:
wherein the cluster number C, m =2 is a smoothing parameter, the gaussian kernel function K (x, y), γ is a weighting factor of the spatial information, and m is 0 Is a set threshold.
4. The method for kernel-intuitive fuzzy clustering image segmentation based on differential mutation grayish wolf optimization according to claim 1, wherein the determination method in the step 5 is:
(1) When rand1<1-t/MaxDT，According to
Updating the position of the gray wolf, X i1 、X i2 、X i3 Is the position vector of any 3 grey wolfs in the wolf group, i1, i2, i3 are belonged to [1, N]And are not repeated with each other,
is a random scaling factor; t is a specific time, X i (t + 1) is the current position of the gray wolf at time t +1, X i1 (t) is the position vector of the 1 st gray wolf in the wolf group at the time t,
the random scaling quantity is the difference value of the position vectors of the 2 nd grey wolf and the 3 rd grey wolf in the wolf group at the time t;
(2) When 1-t/MaxDT is less than or equal to rand1 is less than 1, according to
The position of the gray wolf is updated,
A＝(2×r 1 -1)×a、C＝2×r 2 a =2-2t/MaxDT, A, C is the coefficient vector, a is the linear convergence factor, linearly decreasing from 2-0 with the number of iterations, r 1 、r 2 Is [0, 1 ]]MaxDT is the maximum number of iterations.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110242352.8A CN112767432B (en) | 2021-02-24 | 2021-02-24 | Nuclear intuition fuzzy clustering image segmentation method based on differential mutation grayish wolf optimization |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110242352.8A CN112767432B (en) | 2021-02-24 | 2021-02-24 | Nuclear intuition fuzzy clustering image segmentation method based on differential mutation grayish wolf optimization |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112767432A CN112767432A (en) | 2021-05-07 |
| CN112767432B true CN112767432B (en) | 2022-10-25 |
Family
ID=75691347
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110242352.8A Active CN112767432B (en) | 2021-02-24 | 2021-02-24 | Nuclear intuition fuzzy clustering image segmentation method based on differential mutation grayish wolf optimization |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112767432B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113688960A (en) * | 2021-10-27 | 2021-11-23 | 南昌工程学院 | Grey wolf optimization GHFCM-based residential power data clustering method and device |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109145921B (en) * | 2018-08-29 | 2021-04-09 | 江南大学 | Image segmentation method based on improved intuitive fuzzy C-means clustering |
| CN109886976A (en) * | 2019-02-19 | 2019-06-14 | 湖北工业大学 | A kind of image partition method and system based on grey wolf optimization algorithm |
| CN112085059B (en) * | 2020-08-06 | 2023-10-20 | 温州大学 | Breast cancer image feature selection method based on improved sine and cosine optimization algorithm |
| CN111932556A (en) * | 2020-08-11 | 2020-11-13 | 中国科学院微小卫星创新研究院 | Multilevel threshold image segmentation method |
| CN112163570B (en) * | 2020-10-29 | 2021-10-19 | 南昌大学 | SVM (support vector machine) electrocardiosignal identification method based on improved Husky algorithm optimization |
-
2021
- 2021-02-24 CN CN202110242352.8A patent/CN112767432B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN112767432A (en) | 2021-05-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108596258B (en) | Image classification method based on convolutional neural network random pooling | |
| CN110189334B (en) | Medical image segmentation method of residual error type full convolution neural network based on attention mechanism | |
| CN109800754B (en) | Ancient font classification method based on convolutional neural network | |
| CN110048827B (en) | Class template attack method based on deep learning convolutional neural network | |
| CN113313657B (en) | Unsupervised learning method and system for low-illumination image enhancement | |
| CN109711426B (en) | Pathological image classification device and method based on GAN and transfer learning | |
| CN109754078A (en) | Method for optimization neural network | |
| JP5025893B2 (en) | Information processing apparatus and method, recording medium, and program | |
| WO2019136772A1 (en) | Blurred image restoration method, apparatus and device, and storage medium | |
| CN107392919B (en) | Adaptive genetic algorithm-based gray threshold acquisition method and image segmentation method | |
| CN115661144B (en) | Adaptive medical image segmentation method based on deformable U-Net | |
| CN111652321A (en) | Offshore ship detection method based on improved YOLOV3 algorithm | |
| CN111242906B (en) | Support vector data description breast image anomaly detection method | |
| CN114511576B (en) | Image segmentation method and system of scale self-adaptive feature enhanced deep neural network | |
| CN111784595B (en) | Dynamic tag smooth weighting loss method and device based on historical record | |
| US20240054345A1 (en) | Framework for Learning to Transfer Learn | |
| CN108010069A (en) | Optimize the rapid image matching method of algorithm and grey correlation analysis based on whale | |
| WO2010043954A1 (en) | Method, apparatus and computer program product for providing pattern detection with unknown noise levels | |
| CN112767432B (en) | Nuclear intuition fuzzy clustering image segmentation method based on differential mutation grayish wolf optimization | |
| Zhang et al. | Broadening differential privacy for deep learning against model inversion attacks | |
| CN107240100B (en) | Image segmentation method and system based on genetic algorithm | |
| CN116563682A (en) | Attention scheme and strip convolution semantic line detection method based on depth Hough network | |
| Wei et al. | An incremental self-labeling strategy for semi-supervised deep learning based on generative adversarial networks | |
| Chen et al. | DOF: A demand-oriented framework for image denoising | |
| CN115293966A (en) | Face image reconstruction method and device 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |