1. An image segmentation method based on strong and weak union semi-supervised intuitionistic fuzzy clustering is characterized by comprising the following steps:

(1) inputting an image X to be segmented, and setting initial parameter values: the number k of clusters, the maximum number of iterations T is 100, and the termination threshold e is 10^-5；

(2) Manually marking an image X to be segmented to obtain manual prior information;

(3) the image X to be divided is subjected to intuitive fuzzification processing to solve each pixel point X of the image_jCorresponding degree of membership mu (x)_j) Non-membership degree v (x)_j) Angle of hesitation pi (x)_j)；

(4) Dividing an image X to be segmented into Q different sub-regions R ═ { R) by using SLIC algorithm₁,R₂,…,R_i,…,R_QIn which R is_iRepresenting the ith sub-region, within each sub-regionThe pixels all have different degrees of similarity;

(5) designing a strong and weak combined semi-supervised strategy for class label transmission, and solving the strong supervision membership degree of an image by using the prior information of artificial markersDegree of membership under weak supervisionAnd initial intuitive fuzzy clustering center

(5a) Using artificially marked pixels as strong labels Y^SAssigning the same class label as the strong label to all pixels in the super-pixel region where the strong label is located, and using the class label as the weak label Y after the region label propagation^WThen strong label Y^SAnd weak label Y^WRespectively converted into strong prior membershipAnd weak prior degree of membership

(5b) Using strong a priori membershipAnd weak prior degree of membershipEstimating membership degree of unmarked pixels, and calculating to obtain strong estimation membership degreeSum weak estimate membership

(5c) Respectively estimating the strong degree of membershipSum weak estimate membershipStrong prior membership to eachAnd weak prior degree of membershipMerging as strong supervised membership after class label transmissionAnd degree of membership in weak supervision

(5d) Degree of membership of weak supervisionBringing inCalculating an initial clustering center c_i(1) Then, the initial intuitive fuzzy clustering center is obtained by performing intuitive fuzzification processing on the initial intuitive fuzzy clustering center

(6) Introducing kernel function, strong supervision membership degree and weak supervision membership degree into intuitive fuzzy clustering target function, and designing strong and weak combined semi-supervised intuitive fuzzy clustering target function J_LP-SKIFCM：

Wherein the content of the first and second substances,an intuitive fuzzy set representation representing a color image having N pixel points,is the jth pixel x_jIs the number of clusters, u_ijRepresenting a pixel x_jFor the membership degree of the ith class, satisfy Intuitive fuzzy clustering center, μ (c), representing class i_i) Representing the center of the cluster c_iCorresponding degree of membership, v (c)_i) Representing the center of the cluster c_iCorresponding non-membership, π (c)_i) Representing the center of the cluster c_iCorresponding degree of hesitation, η₁Is a weight index, η, of a strong supervision term₂Is a weight index of the weakly supervised term,representing the strong supervision membership degree of the jth pixel point to the ith class,representing a pixel x_jFor weak supervised membership of class i,an intuitive fuzzy distance metric representing the introduced kernel function;

(7) minimizing an objective function J using a Lagrange multiplier method_LP-SKIFCMTo find out the degree of membership u_ijAnd intuitive fuzzy clustering centerAnd iteratively calculating the membership u according to the updated formula_ijAnd intuitive fuzzy clustering center

(8) Judging an iteration termination condition: if it isOr the iteration times T is more than T, the membership matrix U and the intuitive fuzzy clustering center are obtainedExecuting (9); otherwise, let t be t +1, return iteration and calculate membership u again according to the updated formula_ijAnd intuitive fuzzy clustering center

(9) And classifying the obtained membership matrix U according to a maximum membership principle to obtain a clustering label of the image pixel, and outputting a segmentation result of the image X.

2. The method according to claim 1, wherein the step (3) of determining each pixel point x of the image_jCorresponding degree of membership mu (x)_j) The formula is as follows:

μ(x_j)＝(μ^R(x_j),μ^G(x_j),μ^B(x_j))，

wherein, mu^R(x_j) For pixel point x in a colour image_jThe membership degree under the R channel is calculated by using a maximum and minimum normalization method, andrespectively representing the maximum value and the minimum value of the image X under the R component;

μ^G(x_j) For pixel point x in a colour image_jDegree of membership under G channel, use thereofThe calculation is carried out according to the calculation,andrespectively representing the maximum value and the minimum value of the image X under the G component;

μ^B(x_j) For pixel point x in a colour image_jDegree of membership under B channel, use thereofThe calculation is carried out according to the calculation,andrespectively representing the maximum and minimum values of image X under the B component.

3. The method according to claim 1, wherein the step (3) of determining each pixel point x of the image_jCorresponding non-membership degree v (x)_j) And a degree of hesitation pi (x)_j) The method is solved by utilizing a Segno intuitive fuzzy generation operator, and the formulas are respectively as follows:

π(x_j)＝1-μ(x_j)-v(x_j)，

wherein, delta is a variable parameter, and the value range is (-1, infinity).

4. The method according to claim 1, wherein the strong label Y is used in (5a)^SConversion to strong prior membershipTwo different pixel transformations were included:

for pixels x without strong labels_uWith a corresponding degree of membership of 0, i.e.Wherein the content of the first and second substances,is a pixel x without a strong label_uFor strong prior membership of class i, i ∈ {1,2, …, k };

for strongly labeled pixels x_lAnd belong to the i-th class, thenIf not, then,wherein the content of the first and second substances,for strongly labelled pixels x_lFor strong a priori membership of class i,for strongly labelled pixels x_lStrong for class tPrior membership, te {1,2, …, k, t ≠ i }.

5. The method according to claim 1, wherein in (5a), the weak label Y is labeled^WConversion to weak prior membershipTwo different pixel transformations were included:

for pixel x 'without weak label'_uWith a corresponding degree of membership of 0, i.e.Wherein the content of the first and second substances,is pixel x 'without weak label'_uFor weak prior membership of class i, i ∈ {1,2, …, k };

for pixel x 'with weak label'_lAnd belong to the i-th class, thenIf not, then,wherein the content of the first and second substances,is a pixel x 'with a weak label'_lFor weak a priori membership of class i,is a pixel x 'with a weak label'_lFor weak prior membership of class t, t ∈ {1,2, …, k, t ≠ i }.

6. The method of claim 1, wherein (5b) uses strong a priori membershipEvaluating strength to estimate membershipThe formula is as follows:

wherein the content of the first and second substances,for strongly labelled pixels x_lFor strong a priori membership of class i,pixel x without strong mark_uFor strong estimated membership of class i,SL denotes a strongly labeled set of pixels,indicating a strongly marked pixel x_lAnd pixels x without strong marks_uThe euclidean distance between.

7. The method of claim 1, wherein weak a priori membership is used in (5b)Weak estimation membershipThe formula is as follows:

wherein the content of the first and second substances,is a pixel x 'with a weak label'_lFor weak a priori membership of class i,weak Mark-free Pixel x'_uFor weak estimated membership of class i,WL denotes the set of pixels with weak labels,indicates a weakly labeled pixel x'_lAnd pixel x 'without weak mark'_uThe euclidean distance between.

8. The method according to claim 1, wherein the strong and weak joint semi-supervised intuitive fuzzy clustering objective function J of (6)_LP-SKIFCMIn (1), introducing an intuitive fuzzy distance metric of kernel functionThe definition is as follows:

wherein the content of the first and second substances,is a gaussian radial basis function, and the formula is as follows:

wherein σ represents a scale parameter of the kernel function, and the calculation formula is: denotes the jth pixel x_jTo the ith cluster center c_iThe formula is as follows:

9. the method according to claim 1, wherein the degree of membership u in (7)_ijIs represented as follows:

10. the method of claim 1, wherein the intuitive fuzzy clustering center in (7)Respectively, as follows:

wherein the content of the first and second substances,is a pixel x_jTo the clustering center c_iThe kernel metric at the degree of membership,is a pixel x_jTo the clustering center c_iThe kernel metric at the non-membership level,is a pixel x_jTo the clustering center c_iNuclear metric at hesitation.

Background

Image segmentation is a pivotal link between image processing and subsequent image understanding, and is always a hot problem for researches of scholars, and the image segmentation occupies an increasingly important position. The purpose of image segmentation is to divide an image into a plurality of sub-regions with different attributes and without intersection according to the characteristics of the image, each pixel in each sub-region has similar characteristics of different degrees, and the pixel characteristics of different sub-regions have obvious difference. In recent years, image segmentation techniques have provided reliable and effective help in the fields of satellite remote sensing, smart security, unmanned driving, medical image processing, and biometric identification. In the practical application process, with the increasingly complex segmentation scenes, the performance requirements of people on the image segmentation technology are more and more strict, and segmentation algorithms based on thresholds, regions, clusters, edges and artificial neural networks are developed successively. The clustering-based image segmentation algorithm has the advantages of low computational complexity, good algorithm stability, high running speed and the like, and is generally concerned by students. The common clustering method mainly comprises a hard clustering algorithm, a fuzzy clustering algorithm, a hierarchical clustering algorithm, a density peak value clustering algorithm, a spectral clustering algorithm and the like. The fuzzy clustering algorithm is based on the basic idea of the fuzzy set theory, gives the membership of each sample point data to different classes, can closely represent things and things in the objective world, and is widely concerned by students.

Liujianzhuang in 1992 proposed an image fuzzy clustering segmentation method based on a two-dimensional histogram, which is an unsupervised clustering method based on local search, considers the gray information of pixel points and the spatial related information of the pixel points and the neighborhood thereof, constructs a fuzzy C-mean clustering objective function by using classical Euclidean distance, obtains the membership degree of the pixel points through iterative computation, and realizes image segmentation by the membership degree of each pixel point. The method has two problems in the realization of image segmentation: firstly, a small amount of prior information which can be obtained manually is not utilized, so that the optimal solution is searched blindly and is easy to fall into local optimization, and the image segmentation performance of the background distribution is not ideal; the second is that much ambiguity and uncertainty in the image is not taken into account, making the segmentation of some blurred pixels inaccurate. Aiming at the first problem, Yasunori et al propose in 2009 to introduce the supervision membership degree into the fuzzy C-means clustering algorithm, construct a semi-supervised fuzzy C-means clustering algorithm, and guide the clustering process by using a small amount of supervision information, thereby improving the clustering segmentation precision. Aiming at the second problem, Chaira et al find that the introduction of an intuitive fuzzy set theory can consider more fuzziness of data, so that the classification of fuzzy data is more accurate, and provide an intuitive fuzzy clustering method based on an intuitive fuzzy set.

However, in both of the above methods, the fuzzy clustering objective function is constructed by using the classical euclidean distance, only the case of linearly separable data is considered, and actually, in most image segmentation problems, the data to be processed is often linearly inseparable, so that it is unreasonable to construct the fuzzy clustering objective function by using the classical euclidean distance. In order to handle the linear inseparable situation in image segmentation, scholars also propose a method for introducing a kernel function, which transforms linear inseparable data in an original space into a feature space with higher dimensionality, and finds a linear function in the feature space with higher dimensionality to realize data segmentation. In 2012, Li et al proposed a semi-supervised kernel fuzzy C-means data clustering algorithm based on proximity, which effectively combines semi-supervision and KFCM algorithms to divide linear inseparable data, and guides clustering by using proximity between user input data, and finally verifies feasibility and superiority of the method through simulation experiments on synthetic data. However, the method still does not consider more ambiguity of data and does not fully utilize manual prior information, so that the method has the problems of sensitivity to an initial value, easy falling into a local optimal solution and non-ideal image segmentation performance for uneven background distribution.

Disclosure of Invention

The invention aims to provide an image segmentation method based on strong and weak joint semi-supervised intuitive fuzzy clustering to reduce the sensitivity to an initial value, avoid falling into local optimization, realize the segmentation of low-dimensional linear irreparable data and improve the image segmentation accuracy of uneven background distribution aiming at the defects of the prior art.

To achieve the above object, the technique of the present invention comprises:

(1) inputting an image X to be segmented, and setting initial parameter values: the number k of clusters, the maximum number of iterations T is 100, and the termination threshold e is 10^-5；

(2) Manually marking an image X to be segmented to obtain manual prior information;

(3) the image X to be divided is subjected to intuitive fuzzification processing to solve each pixel point X of the image_jCorresponding degree of membership mu (x)_j) Non-membership degree v (x)_j) Angle of hesitation pi (x)_j)；

(4) Dividing an image X to be segmented into Q different sub-regions R ═ { R) by using SLIC algorithm₁,R₂,…,R_i,…,R_QIn which R is_iRepresenting the ith sub-region, wherein the pixels in each sub-region have different degrees of similarity;

(5) designing a strong and weak combined semi-supervised strategy for class label transmission, and solving the strong supervision membership degree of an image by using the prior information of artificial markersDegree of membership under weak supervisionAnd initial intuitive fuzzy clustering center

(5a) Using the manually marked pixels as strong labels YS, giving the same category labels as the strong labels to all the pixels in the super-pixel region where the strong labels are located, and using the category labels as weak labels Y after the regional labels are spread^WThen strong label Y^SAnd weak label Y^WRespectively converted into strong prior membershipAnd weak prior degree of membership

(5b) Using strong a priori membershipAnd weak prior degree of membershipEstimating membership degree of unmarked pixels, and calculating to obtain strong estimation membership degreeSum weak estimate membership

(5c) Respectively estimating the strong degree of membershipSum weak estimate membershipStrong prior membership to eachAnd weak prior degree of membershipMerging as strong supervised membership after class label transmissionAnd degree of membership in weak supervision

(5d) Degree of membership of weak supervisionBringing inCalculating an initial clustering center c_i(1) Then, the initial intuitive fuzzy clustering center is obtained by performing intuitive fuzzification processing on the initial intuitive fuzzy clustering center

(6) Introducing kernel function, strong supervision membership degree and weak supervision membership degree into intuitive fuzzy clustering target function, and designing strong and weak combined semi-supervised intuitive fuzzy clustering target function J_LP-SKIFCM：

Wherein the content of the first and second substances,an intuitive fuzzy set representation representing a color image having N pixel points,is the jth pixel x_jK is the number of clusters, u_ijRepresenting a pixel x_jFor the membership degree of the ith class, satisfy Intuitive fuzzy clustering center, μ (c), representing class i_i) Representing the center of the cluster c_iCorresponding degree of membership, v (c)_i) Representing the center of the cluster c_iCorresponding non-membership, π (c)_i) Representing the center of the cluster c_iCorresponding degree of hesitation, η₁Is a weight index, η, of a strong supervision term₂Is a weight index of the weakly supervised term,represents the jth pixelPoints strong supervised membership to class i,representing a pixel x_jFor weak supervised membership of class i,an intuitive fuzzy distance metric representing the introduced kernel function;

(7) minimizing an objective function J using a Lagrange multiplier method_LP-SKIFCMTo find out the degree of membership u_ijAnd intuitive fuzzy clustering centerAnd iteratively calculating the membership u according to the updated formula_ijAnd intuitive fuzzy clustering center

(8) Judging an iteration termination condition: if it isOr the iteration times T is more than T, the membership matrix U and the intuitive fuzzy clustering center are obtainedExecuting (9); otherwise, let t be t +1, return iteration and calculate membership u again according to the updated formula_ijAnd intuitive fuzzy clustering center

(9) And classifying each pixel point according to the maximum membership principle by using the obtained membership matrix U to obtain a clustering label of the image pixel, and outputting a segmentation result of the image X.

Compared with the prior art, the invention has the following beneficial technical effects:

firstly, the invention designs a strong and weak combined semi-supervised strategy of class label transmission, fully utilizes the prior information which can be obtained manually, leads the prior information to effectively guide the clustering process, and solves the problems that an intuitionistic fuzzy clustering algorithm is sensitive to an initial value and is easy to fall into local optimum.

Secondly, the kernel function is introduced into the intuitive fuzzy clustering algorithm, so that the linear inseparable condition when the intuitive fuzzy clustering algorithm is applied to image segmentation is effectively processed.

Thirdly, the method utilizes the kernel function, the strong supervision membership degree and the weak supervision membership degree to construct the fuzzy clustering target function based on the strong and weak union semi-supervision intuition, improves the searching performance and the searching optimization performance and enables the segmentation effect to be more ideal.

Drawings

FIG. 1 is a flow chart of an implementation of the present invention;

FIG. 2 is a comparison graph of the results of simulation segmentation of images numbered 124084 in a Berkeley image database using the present invention and existing methods;

FIG. 3 is a comparison of results of simulation segmentation of a Nopeeking number image in a Weizmann image database using the present invention and a prior art method.

Detailed Description

The following detailed description of the embodiments and effects of the invention is provided in conjunction with the accompanying drawings:

referring to fig. 1, the implementation steps of the present invention include the following:

step 1: inputting an image X to be segmented, and setting an initial parameter value and a manual scribing mark.

1.1) inputting an image X to be segmented, setting the clustering number k, setting the maximum iteration number T to be 100, and setting the termination threshold epsilon to be 10^-5；

1.2) carrying out manual marking on each class on the image to be segmented according to the class number k to be segmented to obtain manual prior information.

Step 2: the image X to be divided is subjected to intuitive fuzzification processing to solve each pixel point X of the image_jCorresponding degree of membership mu (x)_j) Non-membership degree v (x)_j) Angle of hesitation pi (x)_j)。

2.1) solving each pixel point x of the image_jCorresponding toDegree of membership mu (x)_j) The formula is as follows:

μ(x_j)＝(μ^R(x_j),μ^G(x_j),μ^B(x_j))，

wherein, mu^R(x_j) For pixel point x in a colour image_jThe membership degree under the R channel is calculated by using a maximum and minimum normalization method, andrespectively representing the maximum value and the minimum value of the image X under the R component;

μ^G(x_j) For pixel point x in a colour image_jDegree of membership under G channel, use thereofThe calculation is carried out according to the calculation,andrespectively representing the maximum value and the minimum value of the image X under the G component;

μ^B(x_j) For pixel point x in a colour image_jDegree of membership under B channel, use thereofThe calculation is carried out according to the calculation,andrespectively representing the maximum and maximum values of image X under B componentA small value;

2.2) solving each pixel point x of the image by utilizing a Segno intuitive fuzzy generation operator_jCorresponding non-membership degree v (x)_j) And a degree of hesitation pi (x)_j)：

π(x_j)＝1-μ(x_j)-v(x_j)，

Wherein, delta is a variable parameter, and the value range is (-1, infinity).

And step 3: and dividing the region of the image X to be segmented by using an SLIC algorithm.

Dividing an image X to be segmented into Q different sub-regions R ═ { R) by using SLIC algorithm₁,R₂,…,R_i,…,R_QIn which R is_iRepresenting the ith sub-region, with pixels within each sub-region having different degrees of similarity.

And 4, step 4: designing a strong and weak combined semi-supervised strategy for class label transmission, and solving the strong supervision membership degree of an image by using the prior information of artificial markersDegree of membership under weak supervisionAnd initial intuitive fuzzy clustering center

4.1) artificially marked pixels as Strong labels Y^SFor strong label Y^SAll pixels in the super pixel region are assigned with the same class label as the strong label and are used as the weak label Y after the region label is propagated^WThen strong label Y^SAnd weak label Y^WRespectively converted into strong prior membershipAnd weak prior degree of membership

4.1.1) Strong tag Y^SConversion to strong prior membership by two different pixels

For pixels x without strong labels_uWith a corresponding degree of membership of 0, i.e.Wherein the content of the first and second substances,is a pixel x without a strong label_uFor strong prior membership of class i, i ∈ {1,2, …, k };

for strongly labeled pixels x_lAnd belong to the i-th class, thenIf not, then,wherein the content of the first and second substances,for strongly labelled pixels x_lFor strong a priori membership of class i,for strongly labelled pixels x_lFor the strong prior membership degree of the t-th class, t belongs to {1,2, …, k, t ≠ i };

4.1.2) Weak tag Y^WThe two different pixels are converted into weak prior membership degrees as follows

For pixel x 'without weak label'_uDegree of membership corresponding theretoIs 0, i.e.Wherein the content of the first and second substances,is pixel x 'without weak label'_uFor weak prior membership of class i, i ∈ {1,2, …, k };

for pixel x 'with weak label'_lAnd belong to the i-th class, thenIf not, then,wherein the content of the first and second substances,is a pixel x 'with a weak label'_lFor weak a priori membership of class i,is a pixel x 'with a weak label'_lFor weak prior membership degree of the t-th class, t belongs to {1,2, …, k, t ≠ i };

4.2) Using Strong Prior membershipAnd weak prior degree of membershipEstimating membership degree of unmarked pixels, and calculating to obtain strong estimation membership degreeSum weak estimate membership

4.2.1) Using Strong Prior membershipEvaluating strength to estimate membership

Wherein the content of the first and second substances,for strongly labelled pixels x_lFor strong a priori membership of class i,pixel x without strong mark_uFor strong estimated membership of class i,l ∈ SL, SL denotes a strongly labeled set of pixels,indicating a strongly marked pixel x_lAnd pixels x without strong marks_uThe euclidean distance between;

4.2.2) Using Weak prior membershipWeak estimation membership

Wherein the content of the first and second substances,is a pixel x 'with a weak label'_lFor class iThe weak a priori degree of membership of,is a pixel x 'without a weak mark'_uFor weak estimated membership of class i,l ∈ WL, WL denotes the set of pixels with weak labels,indicates a weakly labeled pixel x'_lAnd pixel x 'without weak mark'_uThe euclidean distance between;

4.3) separately estimating the degree of membership stronglySum weak estimate membershipStrong prior membership to eachAnd weak prior degree of membershipMerging as strong supervised membership after class label transmissionAnd degree of membership in weak supervision

4.4) Using Weak supervision membershipCalculating an initial clustering center c_i(1)：

4.5) to initial clustering center c_i(1) Performing intuition fuzzification processing to obtain an initial intuition fuzzy clustering center

And 5: constructing strong and weak combined semi-supervised intuitive fuzzy clustering target function J_LP-SKIFCM。

5.1) define the kernel function k (x, y) as a Gaussian kernel, which is expressed as:

wherein the content of the first and second substances,σ is a scale parameter, controlling the radial range of action;

5.2) defining an intuitive fuzzy clustering objective function J_IFCMComprises the following steps:

wherein the content of the first and second substances,is a pixel x_jK is the number of clusters, N is the number of data, u_ijRepresenting a pixel x_jFor the membership of the ith class, m is the fuzzy index,cluster center c representing class i_iIntuition model

The expression of the fuzzy set is carried out,is thatAndthe intuitive euclidean distance between, expressed as:

5.3) Kernel function k (x, y), Strong supervision membershipDegree of membership under weak supervisionIntroduced into the objective function J of intuitive fuzzy clustering_IFCMIn the method, a strong and weak combined semi-supervised intuitive fuzzy clustering target function J is obtained_LP-SKIFCM：

Wherein the content of the first and second substances,an intuitive fuzzy set representation representing a color image having N pixel points,is the jth pixel x_jK is the number of clusters, u_ijRepresenting a pixel x_jFor the membership degree of the ith class, satisfy Intuitive fuzzy clustering center, μ (c), representing class i_i) Representing the center of the cluster c_iCorresponding degree of membership, v (c)_i) Representing clustersCenter c_iCorresponding non-membership, π (c)_i) Representing the center of the cluster c_iCorresponding degree of hesitation, η₁Is a weight index, η, of a strong supervision term₂Is a weight index of the weakly supervised term,representing the strong supervision membership degree of the jth pixel point to the ith class,the weak supervision membership degree of the jth pixel point to the ith class is represented,an intuitive fuzzy distance metric, representing the introduction of a kernel function, is defined as follows:is a radial basis function of the gaussian,a scale parameter representing a kernel function.

Step 6: minimizing an objective function J using a Lagrange multiplier method_LP-SKIFCMTo find out the degree of membership u_ijAnd intuitive fuzzy clustering centerIs more recent.

6.1) to the objective function J_LP-SKIFCMTo find out the membership u_ijThe partial derivatives of (a) to obtain an updated formula of the membership degree, which is expressed as follows:

6.2) to the objective function J_LP-SKIFCMRelating to a cluster centerPartial derivatives of (A) to obtainIntuitive fuzzy clustering centerIs expressed as follows:

wherein the content of the first and second substances,is a pixel x_jTo the clustering center c_iThe kernel metric at the degree of membership,

is a pixel x_jTo the clustering center c_iThe kernel metric at the non-membership level,

is a pixel x_jTo the clustering center c_iNuclear metric at hesitation.

And 7: iterative computation of membership u_ijAnd intuitive fuzzy clustering centerObtaining a membership matrix U and an intuitive fuzzy clustering center

7.1) number of initialization iterations t ═ 1

7.2) degree of membership u according to 6.2)_ijAnd intuitive fuzzy clustering centerThe membership degree u under each iteration is calculated in an iterative way_ijAnd intuitive fuzzy clustering center

7.3) calculationAndthe difference of (a):whereinIndicating the intuitive fuzzy cluster center at the t-th iteration,representing the intuitive fuzzy clustering center under the t-1 iteration;

7.4) comparing the difference Z of 7.3) with a termination threshold epsilon, or comparing the iteration time T with the maximum iteration time T, and judging a termination condition:

if Z is less than epsilon or T is more than T, acquiring a membership matrix U and an intuitive fuzzy clustering centerExecuting the step 8;

otherwise, let t be t +1, return to 7.2).

And 8: and outputting the result of the image X segmentation.

And classifying the obtained membership matrix U according to a maximum membership principle, namely, taking a class label corresponding to the maximum value of the membership in each row in the membership matrix U as the class of the position pixel to obtain a cluster label of the whole image and outputting a segmentation result of the image X.

The technical effects of the invention are further explained by combining simulation experiments as follows:

1. simulation conditions are as follows:

the simulation experiment was performed in the software environment of computer Intel (R) core (TM) i5-4258U CPU @2.40GHz 2.10GHz, 8G memory, MATLAB R2019 a.

2. Simulation content:

simulation 1, the image with the number of 124084 in the Berkeley image database is respectively segmented by using the method of the invention and the existing KFCM method, IFCM method, sSFCM method, SSFC-SC method and eSFCM method, and the result is shown in FIG. 2, wherein:

2(a) is an original image of 124084 image;

2(b) is an artificial marker map of 124084 images;

2(c) is an area tag expansion map of 124084 image;

2(d) is a standard segmentation of 124084 images;

2(e) the segmentation result of the 124084 image by using the existing KFCM method;

2(f) is the segmentation result of 124084 image by using the existing sSFCM method;

2(g) is the segmentation result of 124084 image by the existing SSFC-SC method;

2(h) is the segmentation result of the 124084 image by the existing eSCFM method;

2(i) is the result of segmentation of the 124084 image by the method of the invention.

As can be seen from FIG. 2, the method can completely separate the target from the background for the image with non-uniform background distribution, is insensitive to the initial clustering center, and has a segmentation effect obviously superior to that of the conventional KFCM method, IFCM method, sSFCM method, SSFC-SC method and eSFCM method.

Simulation 2, using the present invention and the existing KFCM method, IFCM method, sfcm method, SSFC-SC method, eSFCM method to segment the images numbered as nonpeeking in the Weizmann image database, respectively, the result is shown in fig. 3, where:

3(a) is an original drawing of a nonpeeking image;

3(b) is a standard segmentation map of the nopeking image;

3(c) is a salt-pepper noisy image of the nonpeeking image, and the noise intensity is 0.05;

3(d) a segmentation result of the nonpeeking image by using the conventional KFCM method;

3(e) the segmentation result of the nonpeeking image by using the existing IFCM method;

3(f) is a segmentation result of the nonpeeking image by using the existing sSFCM method;

3(g) is a segmentation result of the nonpeeking image by using the existing SSFC-SC method;

3(h) is a segmentation result of the nonpeeking image by using the existing eSFCM method;

3(i) is the result of segmenting the nonpeeking image by the method of the present invention.

As can be seen from FIG. 3, the method can completely separate the target from the background for the image with non-uniform background distribution, is insensitive to the initial clustering center, and has a segmentation effect obviously superior to that of the conventional KFCM method, IFCM method, sSFCM method, SSFC-SC method and eSFCM method.